Stay Tuned
Stay connected to RWS with the Globally Speaking LinkedIn newsletter.
Search Globally Speaking website
Enter a query of at least three characters.
108: The Importance of Terminology Work in Machine Translation and Search Engine Optimization
In this episode of the Globally Speaking podcast, we talk to Klaus Fleischmann, a leading terminology expert, the CEO of eurocom Translation Services and founder and CEO of Kaleidoscope. For 25 years, Klaus has been in the translation, localization and terminology industry and in this episode, he gives great insights into how to effectively use terminology in artificial intelligence (AI), machine translation (MT) and search engine optimization (SEO) strategies. Tune in to hear all this and more.
Download episode
Show transcript
Transcript: The Importance of Terminology Work in Machine Translation and Search Engine Optimization
| Speaker | Transcript |
|---|---|
| Robert | Hi, I’m Robert Jelenic, Global Marketing Director at RWS Moravia and I’ll be your host today on Globally Speaking. Today, I’m joined by Klaus Fleischmann. He’s the CEO of Eurocom Translation Services and the founder and CEO of Kaleidoscope. Klaus also served on the GALA Board of Directors for four years and lectures at a series of universities in his native Austria on the topic of terminology. Today, we’ll explore some non-conventional applications of terminology management, the first one being search engine optimization and the second being artificial intelligence solutions. But first, we’ll start with an introduction to what terminology management really is, what it isn’t, and the fact that you’re probably doing it today, whether you know it or not. Let’s let Klaus introduce himself. |
| Klaus | I originally studied translation and interpreting, started out in a translation agency, actually, in California. And then when I came back to Austria, I switched sides a little bit, studied technical communication and started, actually, reselling software. So, I was actually one of the very first. Back then, it was still Trados, now SDL Partners, reselling their software and eventually starting some software ourselves. And one of the flagships that we have is actually around terminology. Terminology, interestingly, always has been very dear to my heart and kind of a passion, always going for the underdogs, I guess. |
| Robert | The topic today is terminology. And we’re going to be digging into some kind of nonconventional applications of that. So, before we dive into it, for maybe newbies like myself, or all the other listeners out there who aren’t steeped in the terminology universe, why don’t we lay the groundwork? So, could you define for our audience, quickly, exactly what we mean by terminology and how it maybe differs from glossaries? That’s maybe a bit of synonymous word thrown around quite a bit. |
| Klaus | Yeah. That’s true. I hear that a lot, actually—terminology and glossary being confused. And it might be a little bit academic to say that they’re not the same. It kind of depends, on what the purpose is, why you are collecting terminology or why you are compiling a glossary. Typically, I find that glossaries use more in the, like, localization context and your goal, typically, there is to make sure that your target languages all translate consistently—same things with same things. Whereas terminology is used more when you talk about source language terminology and then when it goes a little bit more into, like, information management, knowledge management, things like that, which terminology—at the end of the day—it really is. And I like to say that terminology really is the foundation for high quality, meaning readable, understandable, clear, concise, correct communication. And as such, I think it’s obvious that it has to start in the source language. I think everybody and quite a few listeners, I would assume, are from the localization sphere. And we are all very aware that if you have inconsistent and not understandable source content, we might be able to clean it up a little bit in localization, but typically, it’s not a good starting point. So, the difference between glossaries and terminology really is that glossaries, very typically, are focused on the words. So, you’ll have, like, a list of, I don’t know, a couple thousand words if you want, and then you have tables next to them with a little bit of metadata. Whereas terminology really focuses on what we call the concept. So, the important thing is to understand first: What are we actually talking about? And what might different names for that thing be in, let’s say, in our company? And the interesting thing there is that that sounds simple, but it very often isn’t, because in a company, there are, let’s say, different views, different divisions. So if you look, like, at an engineer, be it a mechanical engineer or a software engineer, they have a completely different view of that feature or that thing than maybe a marketing person or a localization person, so they will use names that go more towards, “Well, how do I build this? How do I develop it?“—whereas the technical communicators, they will try to think, “Well, what name will I give this thing, so my audience understands what it’s about?” So, it’s very natural in a company that one and the same thing just gets viewed differently, depending on what your background is and what division you come from. And, also, concepts typically evolve over time. So sometimes, you realize, “Oh, we always thought that this and that is actually the same thing, and now we realize it’s actually not. We actually need to break this up into two different concepts,” which, of course, then has consequences for the target languages, etc. Sometimes you say, “We always thought that it was something else, but actually, it’s the same thing, so, let’s join them.” Of course, the preferred terms might also change with time. So, you start out and then the language changes or, I don’t know, you discover that a term is trademarked by your competition, so you need to change your preferred term. So, this changes, and that’s why it’s so important to focus on the concept, because typically that does not change as much as the words that you use for it. That’s the main difference between glossaries and terminologies—is this concept view versus, really, the term view. |
| Robert | Maybe the glossary is where many people start and they realize, “Oh, actually, this is much more complex than just mapping a word to a word.” We need to start saying, “We have a certain concept. It’s this gray box in front of me with a monitor and a keyboard. That’s the concept. We call it a laptop, or we call it a computer, or we call it a notebook,” or whatever that word needs to be. |
| Klaus | No. No. That’s actually right. And a friend of mine, in a presentation we did at LocWorld—he had this bold statement saying that glossaries are a waste of time, which of course, it was supposed to, you know, be just a bold statement that would make people stop and listen. I don’t think they’re a waste of time, because they really are a good basis to start from. But it’s like you say then, typically, they evolve, and people realize, “Oh, but I have synonyms. Now, what do I do? Okay. I enter a second column into the glossary for the second possible German term.” And then you have five synonyms, and you have five and, you know, eventually say, “Oh, but the second synonym is actually allowed, whereas this, the third one, which is not.” So, you need yet another column. And then, at some point, it just ends, and you can’t live with that anymore. Then, you really need to switch to something that manages this a little better, like a database or something like that, or a termbase, which that’s its function. And the other way around, of course, what you can do is if you have a terminology and you have clean and structured metadata in there, you can produce glossaries. So, for example, you can say, “Well, we have this concept.” But let’s assume we have three different kinds of products that all use the same thing, but it’s called something else in each of the products, which is, actually, a lot of companies have that problem, particularly when they merged with some other companies and, you know, different belief systems clashed with each other. And then, you can perfectly produce a glossary. You can say, “Well, now, I need these 250 concepts, and I always need the term for product A.” You know, leave out all the other ones. They’re just going to bother us, particularly if we talk about applications such as MT, etc., which we’re going to get in later on. It just confuses those systems to have them in there. So, yeah. From the more complex and structured terminology, you can definitely extract a more linear one-to-one, if you want, table of a glossary, and that’s typically how you do it. |
| Robert | Well, I hope that laid some of the groundwork for listeners who aren’t experts in terminology. And so, we’re going to jump into some maybe nonconventional use cases. The first one I’d love to dive into is search engine optimization, SEO. As a marketer myself, it’s near and dear to my heart and I can see the seemingly impossible bringing together of a company defining which terms a concept will have attached to it and what I want to rank for, which is really at the mercy of the public, of what people are searching for in search engines. How does a marketer reconcile the idea of a clean, metadata-driven terminology database with search engine optimization? |
| Klaus | We need to start out with the concept. That’s actually what you do in search engine optimization as well. You start out with a concept and then you basically group words. In terminology, we call that a synonym, and in terminology, typically, the idea is we want to nail this down to exactly one term. All the others are forbidden. We want to use one term and only that one term, always. And of course, in SEO, it’s exactly the other way around. So, the technique you use is actually the same thing. You could call it concept orientation or grouped words or something like that, around a particular concept. But of course, the idea or the goal while you do that is exactly the opposite of normal terminology work. So while normally, in terminology work, you want to make sure we use only one term, and it actually goes as far as saying, “If it’s very, very company-specific, then we need to have the preferred term and only that preferred term.” And in SEO, it’s exactly the other way around. You want to find out, well, what are the most generic terms for this? What is it that people are searching for? And then you basically group those words. So, the method is the same. It’s concept orientation. It’s grouping of words. The goal is exactly the opposite. The goal is trying to find as many generic terms as possible that people out there might be using and not telling them that, “No, you’re using the wrong word,” but trying to, you know, intentionally use that word. Of course, what you then do with that is completely different. So, while the method of grouping terms around a concept and also explaining that concept, etc., that’s perfectly terminology work and that’s also why I think this can be very well done inside a termbase as well. But of course, what you do with the words is different. So in normal terminology, academics get a little bit upset when I use the word you translate terminology into different language because, really, what you do is you look for equivalents and then you kind of make sure that the equivalency is the same. And you do a similar thing in SEO. You look for the click rates and for the match rates and things like that, with, you know, just webmaster tools that will tell you, “People are looking this and that, so if you want to rank higher, you need to use these and that words.” And that basically drives your decision, which should be—I don’t like to call it the preferred term, because it’s not really the preferred term—but which one is the highest-ranking keyword? And then, of course, the interesting thing, also, is in normal, quote, unquote, normal terminology work, typically, the idea is that once you have settled on a preferred term, it stays that preferred term. Right? I mean, you don’t want to change this all the time. So, in SEO, you say, “Well, we’ve optimized on this keyword in, I don’t know, five blog posts before. So, we’re going to optimize on a different keyword in this blog post. And then, in five weeks, after another four blog posts, we want to optimize on yet another keyword.” So, the keyword ranking very often depends on this keyword planning, also, or keyword mapping, where you say, “Well, we know there’s this—four, five, six different things people look for,” and we basically want to spread that across the different, whatever, posts, social media posts, blog posts, whatever you write, or different landing pages on websites, whatever you’re trying to optimize. So, I find that intriguing because the method is the same. The goal is completely different. And what I also find really funny is if I go to companies and I ask them, “Well, how are you currently managing these things?” Let’s see if you guess what the most common answer is, how people manage that. |
| Robert | They don’t. |
| Klaus | That can also happen, but if they do manage it, then they use the one and only tool for everything in the world which is— |
| Robert | Excel. |
| Klaus | —Microsoft Excel. That’s it. Yeah. So, keywords get managed in Microsoft Excel, which you know, that’s also where terminology started, so— |
| Robert | Yeah. It’s where many software packages started, actually. |
| Klaus | Yeah. That’s right. |
| Robert | Many products are born in Excel. Yeah. And so I’m wondering, like for companies that have a strong worldview or strong opinion on the fact that this laptop computer should be called a notebook or should be called a laptop or should be called whatever it is and to create a brand awareness for their product and they may have chosen a, you know, a special word for their product. How do you reconcile that with, essentially adopting the keyword that most people are searching for? Because that’s where the traffic is. That’s what I want to optimize for. |
| Klaus | You do have more than only one keyword per post, so what you typically do is you try to optimize the content to be found. But then, of course, you have secondary keywords where you can try to push your—let’s assume it’s a brand name, maybe. I mean, that would be the most obvious case, that you have a brand name for a laptop. Let’s assume you want to push your own word for that. Then you would try to subliminally, you know, push it in there all the time. Then, of course, you again need the classic terminology management that says, “Don’t forget, if you’re writing about this concept, always sneak in this word, because that’s actually the one that we want to use and that we want to get across.” |
| Robert | So, Klaus, I also understand there are some social media implications related to terminology. Could you walk us through your thinking on that? |
| Klaus | If you look at something like Twitter, then they actually don’t talk about keywords. They actually talk about hashtags, which really is the same thing. And, there, also, it’s very interesting and there are some studies and there’s been some presentations that I was also part of where it was actually measured how changing a hashtag actually changes user behavior because that—you can actually measure. Just, really, like SEO, right? I mean, you have Google Analytics and all these analytics tools that allow you to measure how long do people stay on my page? How many people get on my page to start with if I use this versus? So that’s the classical keyword ranking. But, also, how long do they stay? What do they actually read? What triggers them to click and go somewhere else? All these things. And, you can also do that with Twitter hashtags. So, you can see, “Well, if we change the hashtag from this to that, how does that actually impact, you know, how many views I have, how long people stay on my content and how often I get linked to and all these things?” So it’s really interesting because there’s a few things that SEO and terminology gives us for basically the first time ever, because in the past terminology was just a cost factor but it really wasn’t possible to study or, at least not seriously, study the impact of let’s say, good terminology or smart terminology on whatever communication you put out there. Whereas now you can actually measure that because you can measure the, you know, access click rates, time on page, or on Twitter. You can measure how often these things get clicked on. So, that’s kind of interesting and, also, it puts you on a different spot on the map of the company because so far terminology was only cost. You know? That was like— |
| Robert | It’s the magical shift from a cost center to a revenue generator. |
| Klaus | Absolutely. |
| Robert | Yeah. |
| Klaus | Exactly. And now, you can say, “Well, you know, but we changed this, and look at our click rate.” I mean, you could try to do it, but it was very difficult to argue in the past and say, “We’ve improved terminology, so we now have less support calls.” I mean, maybe it’s also because the product got better, so it’s a little bit hard to really prove that. |
| Robert | Or the documentation. |
| Klaus | Or the documentation. |
| Robert | Right. Yeah. Actually, I think it’s where— |
| Klaus | Or users got smarter. Who knows? |
| Robert | That’s fascinating with hashtags. So, I mean, in essence, that’s a branch of your terminology as well, right? Every post you’re posting, someone has to make a decision on what they’re hashing it. And how do you do that, you know? And how do you control it? And how do you measure it? And where do you say, “This is our preferred hashtag”? Is that the kind of thing you’re seeing companies manage inside terminology as well? |
| Klaus | Yeah. Absolutely. Absolutely. Yeah. And I think managing hashtags, again, is a little bit more classic. I think managing hashtags is the more terminology-related thing than actually SEOing, because again if you want to push your own hashtag, you need to be consistent. So there, it’s a little bit more, again, about making sure that everybody who tweets uses actually the same hashtag. So, that’s a little bit more the classic approach of terminology is, you know, this consistency, which really, SEO is not about. But obviously, also, in a hashtag, you do want to use the one that you know people out there are potentially searching for. |
| Robert | Yeah. Exactly, right? So that’s what I’m thinking. I mean, when I’m logging into Twitter, I’m looking at top hashtags. What’s trending? It’s sort of the same, for a marketer, she still has to reconcile the fact that you may have a strong worldview on what something’s called or how we refer to it, but then you’re also at the mercy of what the market is saying it’s called today because it might be called something else tomorrow. So, it’s really fascinating, this dichotomy of you have to use what people are searching for to be found, but then you have to bring them back to how you want to refer to things so that you have a consistent brand voice when referring to your product or service. I think that’s probably a very fine balancing act that marketers have to pull off. |
| Klaus | Yeah. It is. And the interesting thing there also is: A couple of years ago, still, if you were talking to, let’s say, mainstream engineering companies about something like SEOing or thinking about hashtags, they would, like, just look at you with a big question mark on their face and say, “Oh, well, I understand this with B2C products, that we want to be doing this, but B2B, really?” And that has absolutely arrived. So, I can see that, you know, really, really mainstream engineering companies that build machines, where a few years ago they were like, “Why would I want to do SEO?” They are doing it now, because they see that, obviously, the B2B purchasing people, they now do what consumers did five years ago, you know. They say, I think, B2B is about three years behind B2C. This, it used to be 10 years. |
| Robert | So, would it be fair to say that the key takeaway here is terminology management is alive and well in the marketing world for SEO, and the approach seems to be a mix of classical terminology management and keeping your product lines and service lines and the things we talked about named in a consistent way but balancing that with a live view of what’s being searched for so that you can embed keywords from both sides into your content? Because you need to draw the people in, but then you want to be talking with a consistent voice. |
| Klaus | Yeah. I think it is. And, I think similar as with terminology in general, I think companies are not really aware of that they’re doing this. I think SEO people are typically not aware that it might make sense to use terminology approaches for managing that. They might not be aware that it’s called, you know, concept mapping and grouping synonyms. And for them, it’s just an Excel list with keywords. But really, that is terminology. Just like you say, “You cannot not communicate,” I always like to say, “You cannot not have terminology.” So, all of these things, at the end of the day, come down to terminology. But I think in SEO, marketers are even more unaware of what they’re doing there than, let’s say, the typical technical writers were 10 years ago. I think terminology has arrived at technical writing, definitely. That’s part of every curriculum and everybody knows it exists. Many people might not be doing it in a structured process in their company, but they certainly have to do with it. And I think with marketers, it’s a little bit—I don’t want to say worse—but I don’t see it very often, that marketers are aware of what they are doing actually has to do with terminology. |
| Robert | I guess we all are every day, right? I mean, we all have our own termbase in our head, right? You’ll probably always refer to a certain concept with a certain word because that’s your personal preference. |
| Klaus | Yeah. |
| Robert | We bring that to work. Our teams reflect that. Our companies reflect that. I think the difference is, are you actively, consciously managing it, or not? |
| Klaus | Yeah. That’s right. And I mean, the strange thing is that many companies are not actively managing it and not managing it in an efficient way because they think it’s expensive to do that when, really, that just makes it less expensive because once you structure a process and once you have clear responsibility, that makes it a lot more efficient. And even if you manage to write down the result of that process, which then would be an entry in the termbase, you don’t have to discuss it again in three months’ time because you forgot what your discussion was all about. So, yeah. |
| Robert | It becomes expensive when you add the first, next language or the next five and 10 languages, and you realize your source really needs to be tightened up. Yeah. As a marketer, I find that hugely fascinating and probably something I can take to my everyday job. What if we switched gears into another nonconventional or relatively new area of terminology application—that is, artificial intelligence-driven applications? So neural machine learning and so on. How are you seeing trends in the market? What’s going on when people are thinking, “Will terminology apply to MT engines?” |
| Klaus | I think terminologies are something that in deep learning in general, for the first time I think ever outside of our industry, it’s something that people are interested in. Companies now have deep learning consultants and they come in, and it’s actually a question they ask, “Do you have terminology?” I can’t remember any other case in history where somebody actively, other than translators, came in and said, “Do you have terminology?” So that’s very interesting. Again, just like in SEO before, they very often don’t call it terminology. They might call it structured data or something like that. But really, it’s about terminology. And what they are looking for as well is again, this notion of we have concept maps, so we can identify concepts and not necessarily words so that they would understand that term A actually is just a synonym of term B, but they actually mean the same thing. This is super important, getting this semantic information across. So, this goes a little bit into ontologies and taxonomies. And so, taxonomies being more like, you know, classifying things so you can retrieve them. And ontology’s more into, like, enabling the machine to actually understand the meaning or understand how these, let’s say, eight words belong together and that, what the function is between these words. So, that’s a little bit more than just terminology, but terminology obviously is a very big part in that as well. So, you know, I’m always looking for ways to encourage companies to invest in terminology. And I think SEO, for sure, because we can measure things and we can say, “Hey, it’s revenue driver,” so there it’s a good way to get some investment. And I also think AI. I mean, many, many people say, you know, “In order to prepare yourself for AI, what you need to do is you need to gather as much data about everything that you can possibly think of. You might not yet know what you need it for, but start gathering that data.” And definitely, terminology is one of that. So, if you want to be future-proof, or whatever AI application you might come up with, at some point and in some division of your company, terminology will be good. Of course, a very obvious application is machine translation because there, terminology’s not just in the background trying to explain concepts to the machine or to the learning engine, or it basically stays under the hood, if you want to. But in MT, it actually really pops up, because, you know, text is output, from these engines, so obviously terminology is in that as well. It’s, I think, very much the current state of research of how to make terminology work in machine translation engines. So, in the past, or actually still in the present, largely what you do is you try to get a corpus that is terminologically clean. So, in our world, this usually is like a translation memory, for example. So, you try to get translation memories that are terminologically clean. You clean them up manually, even, and you use that to train an engine. In the past and present, we would try to get a clean corpus, and train an engine with that. And I think if I follow it correctly, what the big MT providers are doing is currently a lot of research into how can we actually give terminology to an engine to make it automatically adjust the output? Because, currently, what you can do is you can give terminology to the engine, but it would use it as a post-processing step, so it would just use its normal algorithm to translate something and then make sure that it always translated the actual terms correctly as a post-processing step. And I think that’s where the current research is going into is, well, how can I actually—much like a termbase in human translation, right—you get a hit from the termbase, and you somehow incorporate that into your TM match, etc., and then make one sentence out of it. That’s currently what MT providers are trying to achieve. So, that’s the current, let’s say, state of research. And, of course, if you pass terminology on to an engine, it’s not enough just to give all the terms to the engine and say, “Take these terms.” So, to start with, very simply, please don’t pass on the forbidden terms. You absolutely don’t want the engine to know the forbidden terms because it would obviously think that it should use them. Also, it might not be enough to just pass the terms, in itself, because you might also need to pass some context information and some, you know, typical use cases of that term to the engine. So, it’s quite interesting to think about, “Well, what does that mean for terminology management? Could we somehow improve, upfront, the data that we can then maybe automatically or periodically feed to the engine by adding, for example, concepts?” So, we’re experimenting with that a little bit by saying, “Okay. Well, if a translator corrects something that was output by the machine, even though it knew that term in this particular case, it didn’t get it, so it didn’t use the correct translation. Well, then, there was something in the algorithm that was stronger than what we’re trying to influence by the term. Then, well, maybe we can add that context, somehow, to the termbase that it would be output automatically the next time to avoid that.” So, that’s basically a little bit where the work in progress is at the moment, I think. |
| Robert | Are there areas like that if you’re trying to transition into an MT program, you have an existing termbase, or maybe you don’t have an existing termbase, are there special areas you would concentrate on specifically? |
| Klaus | Very simply: Don’t confuse the engine. So, whereas for the human translator it can, of course, be very beneficial to offer choices, to say, “Well, if you’re currently in this sort of text, and this is the word you want to use, if you’re in that sort of text, never use that term,” for example. You don’t want to do that to an MT engine. So, what you do instead is you export it into different, whatever the particular system calls it, dictionaries, very often it’s called in the context of MT. So, we have another synonym for this—or maybe not really a synonym—glossaries, dictionaries, terminologies or terminologies, the superset, if you want. So, definitely I would also assume that in the long run, the links between the concepts would be interesting for the engines. As I said, currently, they’re not there yet—as at least the commercial ones that allow dictionaries to be integrated, are not there yet—that they say, “Well, we can have, for example, ontologies exported from the termbase into the MT dictionary, so that we can basically follow the paths and figure out, okay, well, if they’re talking about this, but this is more specific, maybe there’s a blind spot in the target language. But then, maybe we can go to the superordinate term and use that instead.” Still better than nothing. As far as I know, they’re not there yet. I’m sure they’re investigating it because that’s the logical next step. So, I think investing into concept mapping—and explaining to the termbase what’s the relation between the individual concepts—that would be beneficial in the future, as it currently already is for other deep learning engines. |
| Robert | I can imagine a specific use case where you might have a term that in one concept is allowed or preferred, and maybe in another concept is forbidden. If you haven’t cleaned up your concept mapping, you’ll confuse the engine pretty quickly, right? Because now it’s being fed parallel information that this term is both forbidden and allowed or even preferred. And then, you know, missing the sort of implicit understanding that a human translator would have and understanding, “Okay. This concept is the one I want to use because of the context in which I’m writing.” If the computer doesn’t know that, you’re sort of dead in the water. |
| Klaus | Yeah. Absolutely. So, what you would do there is you would split it up into different dictionaries. So, in the termbase, the approach would be, okay, well, typically at least we have one central data pool for the entire, let’s say, organization and then within that we have subdivisions. So, if it’s the same concept in terminology, you would definitely always just throw in all the synonyms that have to do with that particular feature and, then, you would add metadata to each term and say, “Well, use this if that, and use that if something else.” Basically, in an MT engine, similar as in authoring tools, really, you would split that out into different dictionaries. So, then you would have a dictionary for context A or content type A or product range A. But again, as you asked me before with the glossaries, this is perfectly something you could automatically generate from a termbase. |
| Robert | Until the day where the AI goes back and adjusts the termbase for you based on its own— |
| Klaus | I hope not. We still want to stay in control, remember. |
| Robert | Exactly. Yeah. Yeah. That really sounds like the cutting edge. So, this is all stuff being developed right now, yeah? |
| Klaus | Yeah. As far as I’m concerned, yes. So, it’s actually interesting, if you look at some of the manufacturers of MT engines who have basically said they are now focusing on glossaries, rather than enabling the users to customize and tune the engines because they see that that’s really where the problem is. And, you also see that if you look at LSPs, for example, trying to implement machine translation, that really becomes the problem, particularly if you try to implement MT for a customer where there’s large translation memories and large term bases. And then the problem of terminology really becomes the number one problem. It’s not even more the post-editing, but most work actually goes into fixing the terminology. And then sometimes a 70 percent match from the translation memory is actually preferred by the translator over a fairly good machine-translation output that uses the wrong term because, you know, if you have all that data already then you’re almost taking a step backwards. Whereas if you apply this for someone where you don’t really have that much data yet, you’re probably a lot faster to reach savings and speed and all of that. |
| Robert | Fascinating. |
| Klaus | Terminology can actually be negative if you already have a lot and you’re used to the benefits of having it. I have a quote from Kirti Vashee, one of the MT gurus, if you want, who said that “Upfront terminology work is the best input you can give to improve the quality of MT output.” So, just to confirm what we essentially also said but used a lot more words for it. If MT or a little broader use of AI is on your roadmap and you’re thinking about that in your company or LSP or wherever you are, then investing in terminology definitely is a good idea. As I said, I think, really, SEO and MT in our industry—those two are really game-changers for terminology, because so far we all knew that if we really want to have consistent, good quality and correct content and translation, we kind of needed terminology. You could argue that it’s actually cheaper to manage terminology than it is to not manage it. But, at the end of the day, you still survived. I mean, if you had a crappy product with good terminology, that’s still more problematic than the other way around. |
| Robert | Yes. Yeah. |
| Klaus | But for SEO, you definitely need terminology because that’s the essence of it, right? Even though it’s not called terminology, but the method is terminology work. And in AI as well. So there, you need your clean, structured data that does not confuse the algorithms. And that, too, really is terminology. So, this is really a game-changer for companies to invest into this. And for LSPs—if you’re an LSP listening—it’s also a good way of convincing the clients that this is something they need to invest in because this is something that, if you talk to a marketeer like you, Robert, they will definitely understand if you talk about, you know, “You can improve your SEO.” And what you’re doing there is actually terminology work, even though you’re not calling it terminology work. Then they will actually understand that. It’s just not called terminology work. And there’s no cost center for it, so we don’t really know how much we’re spending on it, and that’s actually the only thing that changes when you introduce terminology management. You put the cost center there and say, “Now we know how much we spend each year on terminology work.” And it’s probably less than before but now, unfortunately, we know how much it is. |
| Robert | Right. My point of view would be then, I want to measure how much it’s bringing me, right? And I think that’s something you can do, then, when you thoughtfully manage it, especially in the marketing world now. I mean, especially in B2C, but increasingly in B2B, you can measure things to a shocking degree. So, there’s a point to be made that if you manage your elements like hashtags, you have a really clean way of measuring what’s going on after the fact and the delta between before the management and after the management. |
| Klaus | Yeah. Which is also interesting. Yeah. |
| Robert | Well, Klaus, thank you very much for joining us today. I hope you had fun. |
| Klaus | Thank you. |
Guests
Klaus Fleischmann is a leader in the field of translation, localization and terminology. Not only is he the CEO of eurocom Translation Services and the founder and CEO of Kaleidoscope, he also helps shape the voice of the industry by sitting on the board of the Globalization and Localization Association (GALA). Based in Austria, where he also lectures on terminology at several universities, Klaus has spent 25 years in the field helping a wide range of businesses achieve their international growth goals.
Hosts
Robert is a multilingual B2B marketing leader based in Germany. He’s held past and present leadership roles in technology, sales and marketing. He loves tackling important challenges, fixing big problems and developing and driving growth in a collaborative team environment. Robert is active on his blog.
Listen Everywhere
