RWS Chief Language Officer Maria Schnell on the Value of Linguistic Expertise

SlatorPod #119 - RWS Chief Language Officer, Maria Schnell on LXD platform

In this week’s SlatorPod, we are joined by Maria Schnell to discuss her recent appointment as Chief Language Officer at RWS — pointing to the Super Agency’s investment in linguistic expertise.

Maria begins with her journey as a trained translator and how she worked her way through commercial and operational roles at SDL, before the company was acquired by RWS. She then talks about what the new role of Chief Language Officer entails, specifically leading the Language eXperience Delivery (LXD) platform.

She breaks down the role of the LXD platform, where a large, linguistic network supports clients looking to venture into more remote markets and long-tail languages. She also shares the importance of technology in optimizing operations, from translation productivity to quality assurance.

Maria discusses RWS’ approach to machine translation, where human post-editing is necessary if clients want true engagement with their end customers. She also talks about plans to simplify the technological complexity the Super Agency inherited as a result of past acquisitions.

The pod rounds off with Maria’s initiatives for the next year as she continues to nurture localization talent at RWS Campus and analyze language trends in their language labs.

Transcript

Florian: You have one of the most influential roles in the language industry in terms of the language side as a Chief Language Officer of a top-three super agency, so tell us a bit more about your professional background and your journey in the language industry so far. 

Maria: Yes, I studied translation into Spanish and Portuguese in Heidelberg. I started my professional career as a data analyst and consultant in market research and then joined what at the time was SDL in 2006 as a project manager. From there I worked myself through all sorts of commercial and operational roles until I ultimately reached my current role. I have essentially been in charge of translators, project managers, and any other supporting roles: localization engineers, DTP-ers, audio-video engineers. You name it, I have seen them.

Florian: Chief Language Officer is a new role at RWS and was created recently. Previously, was it an EVP role and now it is chief? Can you tell us more about this role? What services and potentially tech do you oversee in that role and also why was this role created in the first place?

Maria: In a nutshell, I am in charge of the Language eXperience Delivery, our operating platform. Simplified, everybody that contributes to localization and the processes and systems that they interact with which ultimately includes an impressive breadth of capabilities that we have as a group, so our 1600 in-house translators and 30,000+ external vendors that we interact with. It includes all of the supporting roles as well, such as localization engineers, DTP-ers, software testers, audio, video production, and all of the associated vendor and quality managers as well. If we walk away from the pragmatic concrete that I have just described, what is new about the role is that ultimately we have noticed in the last couple of years that the market is drastically changing. This has been going on for a while, but it has just become much clearer specifically after or in the context of COVID. Data explosion is a big topic, so there is lots of content to localize and given that there is so much content, our clients are clearly struggling for their consumers’ attention. Our customers are struggling to be relevant which has led to our customers going more and more granular. We have what I tend to call a target language explosion, so when I started working as a project manager of localization in 2006 I had a lot of customers that would just buy European Spanish and think that should be fine for everybody that speaks Spanish. That is not good enough for most of our clients anymore, even if they send it to Spain. We now have a lot more requests for translations, apart from Castellano, also Gallego, Catalan, Basque, for example. That ultimately means that we felt that to meet that demand you cannot use conventional localization capabilities anymore. You need to have the ability to develop and integrate new capabilities as you go because the whole language market has evolved very quickly, specifically in the context of COVID, so to get there we need to actively influence the market and we need to have the ability to incubate those needs. We already have a comparatively large in-house team. We have historically invested in our campus program. We need to make that more visible because that then ultimately allows us to get better reach in some areas where we want to grow more depth and that is why this new role has now essentially been given this more strategic title and more visibility. 

Florian: You mentioned that you have 1600 in-house linguists. This must be the biggest group of internal linguists in the entire industry, so is that key part of the reason why you retain that big of a group of internal linguists? Or is it that they can interact more with clients? What is the role of this very large group of internal translation resources? 

Maria: In a nutshell, they do that and a little bit more. Again, if we think about relevance, it is something that is becoming more important, so the traditional localization quality paradigm is shifting. You can definitely see that. Defined as the topic of how many errors per 1000 words is becoming less relevant for our clients. What matters a lot more to them is what is the customer experience in the target language, thus Language eXperience Delivery. What generates engagement in the target language? What is relevant to the target group? To be able to do that you need to generate dialogue. You need to talk to the client and to the individuals in the target market that understand their target group to make sure that whatever you translate hits that spot and tone that the client ultimately requires for their end customers or their end-users to engage and to feel that this customer experience is relevant to them. That is what is helpful about those in-house translators. On a fairly daily basis, we do have engagement with clients where a client ultimately gives us, what we would call, a traditional marketing translation. They tell us to translate that into the following 85 target languages and the in-house team then ultimately pushes back to the client and says the use case that you are formulating in the source text may be relevant for the source market, but is not that relevant for the target market that you are going after and then we have this dialogue between the client and our linguists on who do you want to reach and where and how do you want to reach them. Which for me is part of the granularity in the target language topic. One of the themes that we see a lot is that our clients are not only going very local now but they are also going subcultural in many ways. We have a lot of client requests where clients expect us to reach expert communities. To be relevant to an expert community, you need to know when most of them have immigrated because their language will to some degree be frozen at that point in time. We had a translation request for Pennsylvania Dutch. You need to understand the history of that community and give feedback. For example, this channel that we are trying to use to reach that community is a channel that is irrelevant to whoever speaks Pennsylvania Dutch, for example, so that is one big part of what they do. Many of our clients are going more and more remote because it is easier to reach remote communities and it has become a lot more relevant for our clients. Again, in the context of COVID, we were translating a lot of pharmacovigilance material, for example, that was required in the context of the COVID vaccine. That was the first vaccination campaign that went truly global very, very fast so, my team insisted that we found a language delivery team there. We had to find loads of Hawaiian linguists to translate pharmaceutical material. It is an island. There are not a lot of people on that island and most of our clients have 24-hour, 48-hour turnaround agreements with their clients. When it comes to pharmacovigilance that can be a real challenge if essentially what you are dealing with is a handful of linguists that are not even remotely close to the requirements of what you need. What in-house linguists then help with is they build a quality consensus with the client and with the supply chain. That is one of the issues with some of those smaller or more immature languages. I mean that in the most respectful of ways. Immature is defined as you cannot study translation into some of those languages, you will study the target language, but you definitely will not study the tools that you need for translation and you do not necessarily have the means to specialize and that is what the in-house translators help with. The other issue is that you do not necessarily have the Académie Française or the Dublin Institute, somebody who says this is what good German looks like, so you have to create a consensus of what that is so that you have at least a starting point to discuss good quality. That is what the in-house linguists help with as well that has become a real superpower that we have always had but truly discovered as clients got more and more remote and we had to ultimately reach communities and build some of the quality consensuses and build a curriculum to develop people towards translation in most of those markets.

LocJobs.com I Recruit Talent. Find Jobs

LocJobs is the new language industry talent hub, where candidates connect to new opportunities and employers find the most qualified professionals in the translation and localization industry. Browse new jobs now.

LocJobs.com I Recruit Talent. Find Jobs

Florian: You also have the automation component, so how do you work with the tech team, the team at Trados, the AI team, and the Language Weaver team. How does that work to make this coordinated whole? 

Maria: The way we want all of our production team members to work is in this concept of augmented translation, so ultimately the linguist sits at the center of all the tools they have available which means that we make it a point of using as much of our technology as is feasible. Sometimes it is just not feasible. If you get a handwritten scan, for example, it is just not going to happen, so we try to work as much without technology as we can and we have built an operating platform already that we are in the process of optimizing in the next couple of years. That allows us to collect a lot of operational data which has become incredibly helpful because it puts us in a position to feed into the development teams. More clients are asking for certain language sets, a language that you should think about automating using AI capability. It helps us ultimately improve our quality assurance capabilities. Some of those languages that are complex need a lot of input on what is happening in QA, so there is a central data feed that helps development continuously improve their products. We are essentially development teams both in AI and in translation productivity. Essentially, very valuable input for continuous optimization of the platform. What we use this for as well is to continuously optimize our operational processes at the end of the day. My team is incredibly data-driven. Data analysis is what we do on a daily basis to understand what is going on and how we have to react because this industry, as we all know, is incredibly fast-paced, so it helps us react a lot faster to emerging trends and it also helps us better understand what are the key things that we need to do to produce predictable quality output, which feels a little bit like the holy grail of localization. If you have lots of data that you can analyze and that you can act upon, which is even better than analyzing only, that is something that becomes possible. 

Florian: Can we dwell a bit on the language combination? You mentioned the challenge of sourcing Hawaiian if that is even a language or Hawaiian linguists.

Maria: You can dispute whether it is a language. There are people that have opinions in both camps, it is a language and is not a language. There definitely is a Hawaiian language Institute that we are working with to identify linguists. What is a linguist? Your and my conventional definition of a linguist does not always apply frankly in most of the markets that exist on the planet, so there is a lot of interaction, teaching, et cetera, that has to happen, and a lot of technology use that has to happen, which is what my teams are supporting us with.

Florian: What other groups of languages are more in demand and getting up that translation, localization, maturity curve? Any particular region or group of languages that you are seeing?

Maria: As a more general trend, I definitely see a lot more granularity in target language sets. When I started there were only about 100 target languages and we are at 268 now. We had a huge jump in the last two years where we went very granular and the amount of source languages is growing aggressively as well. Again, the Spanish versus Castellano, Galician, et cetera, is a big topic that exists everywhere. The other topic that is a clear trend that has been going on for a while arguably is Indic, Southeast Asian, and African languages and for Indic, Southeast Asian, and African languages what you can clearly see is the mobile connectivity revolution that each of those geographies and societies has seen, that opens up new markets and they are in different stages of becoming more mature in the localization paradigm. Pacific languages are not to be underestimated. That one has blown me away the last two years, so Hawaiian, Tongan, et cetera, has been a real headache for me personally in the last two years, to be frank. 

Florian: Why do we not talk a bit about post-editing machine translation? There is this unstoppable transition to PEMT workflows and how do you see this changing the day-to-day work of linguists and also the demands placed on those linguists?

Maria: My team embraced machine translation a couple of years ago fairly aggressively. From my perspective or their perspective machine translation is here to stay and has been here for a while. One of the things that you take away from engaging that much in machine translation is it has its limitations but it is an incredibly powerful enabler. One phrase that my team uses a lot, which I always love, is nobody loves a robot. If clients want true engagement with their end customers you need a human to take the pre-translated content and make it relevant to the end customer of our clients. That is even true for heavily regulated content to be perfectly honest. The whole customer experience in the target language is a lot more relevant and the whole engagement topic is a lot more relevant, so that is my core takeaway. The other core takeaway is when I started working as a project manager in this industry I had an average turnaround time in projects of about 35 working days. Those times are truly over. Six to 12 hours is what we are talking about now. It would not even be physically possible to do that without using technology as an enabler and that is how we use technology. This translator topic comes back again. The translator uses technology to ultimately become more productive and spend most of the time where a human should be spending time, so again, creating relevance or creating compliance. Ultimately understanding what are the regulatory requirements in the target market and making sure that we hit them. 

Florian: You mentioned regulated, is this generally regulated? Do you find it harder to deploy some of these post-edited or automated workflows or does it matter whether it is marketing or high volume text types or in the regulated space?

Maria: Honestly, since neural machine translation, text type, and content type are much less of an issue, there are very few limitations. Things like software localization to this day is not a particularly brilliant idea to use, even neural machine translation engines, because segments are short and there is so much more that you have to think about. Same for localizing subtitles. For the rest, what you may see is differences in productivity, but it is minor differences in productivity, so a lot more responsibility with neural MT than what was historically traditionally possible before. 

Florian: Tell me a bit more about the Language eXperience Delivery. Is this an internal initiative? Or are you communicating this with the clients with the outside world? 

Maria: The mission statement is language delivery with customer experience at its heart. It is ultimately the platform and the people that interact with that platform, so all the people that report to me, translators, localization, engineers, DTP-ers, et cetera, plus essentially the process that they use and the systems and tools that they use which is a machine-first data-driven operating platform, which we talk about to customers as well. The reason why we talk about that is that we have developed that as an internal optimization tool in the last couple of years, but it has become incredibly powerful in client interactions as well and in the interaction with our divisions. We are a support function to our divisions and we provide them with not just delivery, but also data insights. I stare at my production things on a daily basis, multiple times, and so does the whole team. We recognize earlier things like, there is a certain language strength that is starting to crystallize. Why do not you go ahead and talk to your client about a specific language type? All of your competitors are going into those markets, have you thought about that tool? Is that not necessarily relevant except by accident or is this a decision that you have made? Quality trends that we see across specific content types that we bring up that we notice and use in the context of continuous improvement initiatives and that we then share with our divisional teams and the divisional teams then often come back and go, this is a very interesting topic to talk to our clients about as well. I have been talking to my team a lot about the whole relevance versus accuracy topic in the quality paradigm in the last couple of years and at some point in time, one of our divisional colleagues helped me talk about it and said, can I put you in touch with our clients? Yes, we talk about LXD with our clients as well. 

Florian: What are some of the key initiatives you have for 2022 now and going into next year in your new role as Chief Language Officer? 

Maria: One of the topics that is very dear to my heart is continuing to invest in Campus because given the target language challenge that we talked about earlier, that is definitely something that has proven to be incredibly powerful and it is incredibly rewarding as well, so continuing to reach out to all of our university and other academic partners to just make sure that localization is front and center of language requirements. Also the strategic investment into our language labs. Language labs ultimately are the institutionalized production data analysis function that we have in our teams. What they ultimately do is again, analyze language trends and then turn that into continuous improvement initiatives and turn that into themes that we can discuss with some of our clients as well and then automation and simplification. We have grown through acquisition as a group and as a result of that inherent technology complexity that we are now resolving and from a simplification perspective, we are pretty good and engaging with our linguists in-house and external partners. I do think we can do even better. We can be more helpful in curriculum development. We can be more helpful in how we onboard and how we generally engage, so that is one of the topics that I also want to focus on. That is my agenda for 22 and 23. 

SlatorPod – News, Analysis, Guests

The weekly language industry podcast. On Youtube, Apple Podcasts, Spotify, Google Podcasts, and all other major platforms. Subscribe Now.

SlatorPod – News, Analysis, Guests

Florian: When you work with the universities, a lot of times they would ask me, which skills should we focus on? Do we need to give everybody some basic Python skills now so they understand how those engines work? Or should we squarely focus on language and cultural expertise and the rest will be then provided as they enter the industry? What is your take on that? 

Maria: My observation in the last couple of years is a general curiosity. Not being afraid of technology is a big topic and a general curiosity for the industry. The industry is developing fast and you need to be willing to explore different skills. In the grand scheme of things, we are now defacto asking linguists to bring or either learn skills, depending on which market we are in. Sometimes you cannot expect that. We are expecting translators to be able to post-edit. We are expecting translators to be able to leave the source, so to have this whole cultural adaptation, research capabilities, feel confident to be in a dialogue with the client as well, and we are expecting linguists to be willing to bridge the written word to spoken word gap because we are definitely seeing a trend towards more audio-visual content. Even traditional things like user guides have now become a video in many products, so having the flexibility to evolve with the market is one of the cool topics that is becoming more and more relevant. Honestly, having lived with a lot of automation and AI for a while now, there are zero things to be concerned about in the future. There is so much more to do to remain relevant. As long as we evolve with the market.

Florian: I love when you say leave the source text because as a linguist I found it a little close to the source but it takes a lot of brainpower and that is why the machines cannot do it. 

Maria: It also takes a lot of confidence. I understand the target group that you are trying to deal with, I know what you are saying here, and it is just not working. In the context of COVID, one of our customers who produces and sells sports gear noticed that their target group is changing. People still buy sweatpants, but not to go to the gym, so what are they buying them for? One of the universal use cases that they described was home improvement projects. That may be a relevant use case in the US where their customer sits, but I can tell you it is not relevant for Japan because people do not have a garage and the space to do a home improvement project. What the linguist at the time said is, thank you, this is not a relevant use case for that target culture. For that, you need to know your stuff and feel confident enough to walk away because that is literally dumping the source and writing from scratch.