In his eight years in game dev, Nick Citkoŭ has worked on a huge range of projects, from multiplayer online games to global historical strategy. Now he is a localization producer at Paradox Interactive and helps to launch projects in new markets.
We talked to him about what gets in the way of localizing a game, how to assess the effectiveness of localization, and why translators really don’t have to be native speakers.
“I like it when I can improve something myself”
— You started off as a translator for a finance company and then got into game localization. How did you manage to make the jump between the two fields?
— I always wanted to work in game development. I graduated from university, started looking for my first job, and applied at the same time for two positions, one at Wargaming and the other at the auditors Ernst & Young. By the time the Wargaming team sent me a test translation, I was already at the penultimate stage of interviews with Ernst & Young. I decided not to risk it, chose EY, and worked there for two years. Then I saw a vacancy at Wargaming again and thought, “Hmm, I should probably give it another try.” My second attempt was more successful.
— Was it difficult to switch to gaming?
— There was a lot of new stuff to learn—completely different tools, a different approach, different materials. At Ernst & Young, for example, we used translation databases a lot. Translation Memory is a real help when you need to translate a 100-page report, twenty pages of which are set formulas. There’s no point in translating them again, you just pull whole phrases from the database. But you can’t blindly trust databases, and that’s why we would check everything a thousand times over.
Working at Ernst & Young taught me to be hyper-attentive, because a single wrong figure in auditing for a big corporation can be very expensive. That experience has come in really useful. I still check my work with maximum care. I think that sometimes it’s even worth printing out your text so that you can see it on paper—that changes your perspective.
— So, how did you end up moving from Wargaming to Paradox Interactive? Was it a conscious decision or did the opportunity just arise?
— There was one more step before Paradox Interactive—from Wargaming, I moved to the Belarusian company Vizor Games. Most of what they do is mobile games, but I worked in the department developing an experimental game for PC and consoles. I was hired as localization manager and I worked in the company for around a year and a half. By that time, I realized that I wanted to consider other opportunities.
I went on LinkedIn and I saw the vacancy for a localization producer at Paradox. I was familiar with the company from my time at Wargaming, as we were competitors. Wargaming was developing the remaster of Master of Orion, while Paradox brought out Stellaris. We kept an eye on each other to see what the differences and similarities between the games were.
When I was choosing a new job, I thought that localizing Paradox’s projects would be interesting. Their games are huge and really difficult. Now I can tell you I wasn’t wrong—it’s really not easy here, but it is awesome. There are lots of factors that need to be taken into account, and lots of opportunities for optimization. I really like it when I can improve something myself.
— Wargaming and Paradox make quite different games. Does genre have a big influence on the localization process and on your work specifically?
— A very big influence. There are lots of fundamental details in Paradox games—the names of countries, parties, regions, units, etc. And everything works to scripts: when some action happens in the game, all that information is inserted into the text. This process often leads to complications, because these fundamental names have to agree in all languages.
The second issue is the events that happen in the game depending on your actions, or in relatively random order. There are also loads of them, and they use substitution: the values are taken from the game, but the text is more literary here. So the translator has a little more room to maneuver.
— How do you solve the problem of agreement?
— The system is complicated and it doesn’t work perfectly. We rely on the expertise of our localization staff. Our vendors put a lot of effort into teaching translators to customize scripts so that they work in the required languages. Of course, this means extra work for the translators, but for now it’s the only way. Otherwise, the texts would be wooden.
We do get mistakes with custom scripts and agreements, but overall the texts are fairly lively. And we achieve this thanks, among other things, to the work of our main localization partner and several former and current mod makers that help us with localization scripting. Their team does a huge amount of work, and we’re massively grateful to them.
— What languages are the hardest in that respect?
— Of the ones that we support: Polish and Russian. Apart from them—French, Spanish, and German. If, let’s say, the gender of a ruler changes, then we need customization to correct the endings of the verbs and the articles. We do that with the help of a script: the translator has to write the values of the variables in separate files, which are then processed in the game. It’s a really complicated system, and we’re looking for options to optimize all that.
— How can that be done?
— There are two possibilities. The first is to come up with a grammatical engine ourselves for all the languages we are going to support. We’d look at all the problems that could arise in the particular languages: genders, cases, support of plurals, changes to parts of the clause depending on what variable is substituted. And then we’d work out how the engine could solve all that.
The second option is to go for a pre-existing solution. There’s a huge choice. For the most straightforward example, we have the noun “ax” in English, so we write the translation “hache” in French. We note that it’s feminine and singular. In that case, the adjectives and articles that agree with “hache” will automatically be put in the feminine singular form.
Besides, it’s important to look at the particular game. If the game is linear, there aren’t many characters, and you can count the variables on your fingers, then it’s unlikely you’ll need to invent a grammatical engine of your own. However, if the project comprises millions of possibilities and a huge number of alternative plot developments, it’s worth considering.
— You’ve also developed custom tools that have allowed you to cut localization expenses. Can you tell us how that works?
— I didn’t do it all by myself. As part of my job, I’ve been lucky enough to come across some wonderful people in the localization industry. We came up with cool ideas together and had a chance to implement some of them. We’ve often collaborated when someone spoke up about an inconvenience—together, we’ve worked out how to deal with it. That’s often how extra optimizations are devised.
When I was working at Wargaming, we had a project which often published announcements about discounts on specific types of goods. Every time, we worked to the standard plan: we were sent an announcement, we forwarded it to the translators, and we localized it. Over time, it became clear there was a pattern to the announcements—five introductory phrases and a list of goods that changed.
We wondered what the point was in localizing the introductory part every time if we could standardize it so that it was always the same. Then the goods could be taken from the database and the values substituted in different localizations. In other words, the source announcement arrives, the required values are pulled up, the description comes from the database, and everything’s ready for publication. We made sure the localization came to the content, and not the other way around.
“There has to be someone in the high-level management who understands the importance of localization”
— You’ve paid a lot of attention to setting up the processes of localization. Can you tell us what errors are most often made?
— The fundamental mistake is shortsightedness. When a game is being developed, it’s rare for anyone to think about it being localized into other languages. Sometimes, it’s only once the project is 70-80% ready that people realize. By that time, going back to the start and trying to introduce internationalization is really painful. In that case, the localization is bound not to be perfect or even reasonably good. There will obviously be fundamental flaws in it that can’t later be corrected. That’s exactly why lots of people prefer to have nothing to do with it.
The second mistake is that lots of people don’t consider the differences between alphabets. Several times, I’ve come across the problem of autocapitalization in Turkish. They have two “i” letters, one with the dot and one without. So, if you don’t think about it in advance, you’ll end up with the wrong letter in lots of places. That has to be corrected at the level of code and fonts. You get the same thing with different figure and date formats. In an ideal world, those issues should be on your mind before the code is written, because correcting them with text methods later is basically impossible.
— In your opinion, when’s the ideal time to think about localization?
— I’d say at the very start of working on the text and the interface. When you decide what the game’s going to look like and make the first playable. By then, you already have texts and elements of the interface, and that means you can get a localization expert and ask them to tell you what it’s worth paying attention to so that you can avoid problems later with launching into other language markets. Any good loc manager or producer has a guide with a list of important issues for internationalization. It’s very important to keep these in mind during development.
— Let’s say I’m a localization producer. Who do I need to approach to ensure that localization is being considered as early as possible? And how do I convince them?
— It’s a difficult area and it’s probably one of the biggest problems in the localization field. Ideally, the developers contact the loc producer, rather than the other way around. There has to be someone in the high-level management who understands the importance of localization and is ready to stand up for it with you. If everybody thinks it’s fast, simple, and basically of secondary importance, convincing them otherwise will be difficult.
You can try to convince them diplomatically and present information about similar projects. For example, by showing how project X grew after it was localized into other languages and how much revenue it brought. If that method doesn’t work, you can come back to the question when you hit a bump and say, “We did this and now we have this problem. The players have noticed it, and they’re really unhappy about it.” That may be when they realize that localization isn’t as insignificant as they thought before.
But in a perfect world, of course, you don’t end up in that situation. When I start on a project, I try at the very beginning of the work to talk to the leads of all departments connected with localization. That includes UI, narrative design, content design, and sometimes art too, marketing and support—basically, all the departments where there are letters and languages. I try to understand from the conversation how they work and whether or not I can fit in comfortably. Or whether we’ll have to raze everything to the ground and try to build it all up from scratch.
— So, if you manage to convince your colleagues, how do you set up a localization process that will run throughout development?
— Now there are lots of cool tools in localization that allow you to do everything very early on. There are TMS and CAT tools that integrate with the tools for interface design and lots of other game development systems.
At this stage, you can already produce draft localizations and show what the text will look like in different languages. Then the team will understand that here on this button that we’re planning to put ten symbols on, in Chinese there’ll just be one character. So what do we do? After that, you just have to come up with options: to make the button dynamic or limit it according to minimum length.
I was discussing a case like that just recently with my colleagues: what to do with a button that would have four characters on it in Chinese and 52 symbols in German. We came to the decision that it could be scaled up and down to a defined level.
— I’ll ask what’s probably the hardest question for someone working in localization: how do you judge its effectiveness?
— If anybody knows the exact answer to that question, please write to me [laughs]. I’ve spent a lot of time thinking about that. How do you measure the effect of localization on success in isolation? To be honest, I don’t know. You can, of course, say that if there was no localization, fewer people in that market would’ve bought the game and it would’ve brought you less revenue. But how much less, you don’t know.
If a game was launched in a market without localization, and the localization was subsequently added, then we can make an evaluation. We can see that before localization there were 1,000 players in the region, and afterward there are 10,000. And if nothing else changed, then you say that the localization earned 9,000 players. But that doesn’t happen often. Lots of people prefer to launch localizations as part of a content update, and in that case you can’t accurately discern what caused the increase in players.
— What if you do a soft launch without localization and only add it afterward?
— It’s better to launch everything at once. True, we won’t be able to evaluate how much the localization brings us, but at least we won’t lose our player base because we didn’t localize the game into their language in time.
If, after the launch, you get a growth in players in an unexpected region, however, it’s worth adding localization post factum. For example, we launch a game and we see that activity is suddenly on the up in Türkiye, Italy, or Poland. If the game’s not too big and the budget allows, why wouldn’t you do a localization?
— Does it happen with Paradox that you decide to add a new language pair?
— Yes, but for us in particular, adding a new language after a huge amount of DLC is really hard. There are already a million words in the game, and that means a new localization will cost a lot of money and take ages. Is it really worth it? Not always.
“With translators, there are two extremes”
— In your experience, is it better to hire freelance translators or LSP studios?
— At Paradox, we use LSPs and freelancers. Of course, it’s always more convenient to work with a team, especially when your resources are limited. If we have just one producer for four projects, it’s unlikely they’ll be able to manage ten individual translators per project. What they need is one contact person they can communicate with about any issues. That almost certainly means an LSP. Sometimes it happens that the individual translators who’ve been working with us for a long time join an LSP team and coordinate the work on their side.
LSPs are really good for particular tasks — for example, for working on marketing materials that don’t require any super specific knowledge. If we’re talking about localizing a whole game, however, then I prefer working with individual translators. You can collaborate with them directly and onboard them yourself. That way, I know that they’re almost certainly more involved in the process and I can rely on them. However, there are LSPs that go above and beyond for your product so it’s very situational.
— So you mean it’s a question of trust?
— Yes, I think so. Of course, there are risks—the well-known “bus factor.” If someone gets run over by a bus, what next? But in that case, you need to have other translators or localization agencies in mind that can take over in an emergency. That’s not so hard to organize.
— And what makes a good translator for you?
— It’s somebody who knows the language perfectly and has a real feel for it. Furthermore, they understand the cultural specifics of the countries where the language is spoken. For example, we had a Japanese guy who lived in the States for most of his life, but still retained close ties with Japan and was an excellent translator.
Whether or not a translator is a native speaker isn’t really that important. I’m not prejudiced against people who aren’t natives. If you’ve learned a language to proficiency, can explain yourself perfectly in it, write and translate, then you’re welcome. I worked with a narrative designer who was Russian, but wrote amazing lore texts in English. For me, that was yet another confirmation that it’s not essential to be a native. Furthermore, not every native speaker is capable of becoming a good translator.
— Are there any criteria that stop you from choosing a translator?
— Soft skills. I mean, if somebody translates perfectly, but it’s not possible to communicate with them normally, then we’re probably not going to work together. For example, if they’re toxic and don’t know how to ask questions or answer promptly without being passive-aggressive.
— So a good translator shouldn’t be afraid to ask questions?
— With translators, there are two extremes. The first is when you ask questions as soon as they pop into your head. Then you translate six more strings or take a look at the commentary and you find the answer. The other extreme is when you don’t have questions at all. The translator says, “It’s all good, I understand everything, I’ll do it all.” And then they miss all the context, which they should’ve asked about.
For me, the ideal translator spends a reasonable amount of time contemplating the challenge in front of them before involving you, but if they don’t manage, they won’t be shy to ask a question.
— What else is important to be a good translator?
— Another thing is you have to know what tools there are on the market and how to use them. It’s preferable if you have experience working with one or more tools. It’s essential for translators working in game localization. From my experience, I can say that once you know one CAT tool, you can get a hang of any other fairly quickly—a week is more than enough. Furthermore, you need to know where to look up translations, synonyms, and word usage.
And your aggregate general knowledge about a language is very important too. You need to understand how real people would say what you’re trying to express. Essentially, you can find that out in any work of literature. I think it’s very important that a translator is reasonably well-read and has a good feel for language. That’s necessary to ensure their phrases aren’t wooden or calques.
Moreover, you need to stay up-to-date with the contemporary context and keep your finger on the pulse of current affairs. Sometimes you have to avoid sore spots that weren’t all that important a couple of months ago.
— What about how you organize work so that translators always submit on time, but don’t burn out in the process?
— Now we’re trying to diversify work on projects in a way that means there are no overlaps. By which I mean that a translator who is working on two projects doesn’t get batches from both simultaneously. With our LSP, we agree that there are teams on standby, ready for specific projects that they always work on.
As for me, I try to always ask for a roadmap. Then I can estimate when we’ll be receiving batches and share the information with the LSP in advance. That way it’s easier for them to plan the workload for their localization providers.
If we can see that there’s a giant batch coming, then we’ll try to warn them immediately, “Guys, we’ll soon be flooded with work for two months. Get ready and get some other people involved. The word count will be huge.”
“In historical games, the most important thing is neutrality”
— Let’s say we’ve done a localization and even performed LQA, but something went wrong. What’s the right way to approach a debriefing? What do you focus on?
— First of all, there are standard metrics—for example, the number of errors in a specific number of words. Then you need to pay attention to user feedback. If I’ve been sent a report or I can see that there’s some scandal on the Spanish-language forum, I need to find out why people are unhappy.
There’s one more important element, and that’s feedback from the developer. They may tell us we’ve uploaded a file that overwrote the previous one. The previous one was already updated and that lead to problems. If issues like that come back from the development team, you have to take them into account and think about how to avoid them in the future.
— Do player surveys really help? Because you go to Steam Reviews, for example, and you’ll find a lot of unconstructive criticism.
— That’s true [laughs]. But you can formulate the surveys so that players have to write what exactly they don’t like. You split the questions into specific areas: are they happy with the style, the grammar, the spelling, the font? That way, the player can evaluate lots of different linguistic areas. But I’m sure there are some respondents who put ones everywhere.
In theory, you can send a follow-up asking for more detailed information and examples. You can give them some in-game benefits to compensate them for their time. Also, if you can see that everything’s OK in all the other languages, but you’re getting a hundred negative reviews about the style of one, that’s something to take to the editor or LQA tester. You need to find out why it’s happening.
— Are there player communities that are particularly sensitive to localization errors?
— Anyone can make a typo, that’s not really a serious issue. It’s far worse when you develop a game and you don’t understand the cultural specifics of the markets you’re planning to sell it in. If there’s content in the game that’s considered unacceptable in some countries, that’s an enormous problem that can alienate your audience.
We track reviews for evidence of problems like that, so that if something does go wrong, we can correct it as fast as possible. But in most cases, we try to avoid mishaps in advance. Cases when something got into a game that nobody knew or thought about are very rare.
— But how do you avoid that in Paradox games? For example, the Hundred Years’ War is perceived differently in England and in France.
— The main thing is to stay as neutral as possible. By which I mean that you should avoid any value judgments and convey the course of events as accurately as possible. Just the facts. In terms of historical realism, I think that’s the best approach.
You have to bear in mind that in different countries there are different rules about what can be shown in a game. For example, with regard to symbols, flags, and other elements. That definitely needs to be considered.
— What about if a game isn’t very well received when it’s launched, partly because of the localization? Is there any point in correcting errors?
— That depends on sales plans. If you want to support the game further, then you absolutely have to fix the problem areas, otherwise the players will just stop playing. If the game is unreadable and it’s impossible to play it, then people won’t care about DLC and updates.
If the aim was to hype the game and get as much cash as you can in the first week, then you don’t have to correct anything. You just say, “Sorry, we’ll definitely take your concerns into account in the future, thank you for your constructive feedback.” I mean, I’m not a fan of that approach, of course. It looks like a scam.
I want everything to look good. All deficiencies should be put right. At Paradox, we pay a lot of attention to community feedback, not just in regard to localization but also with events and scripts. We work with the player feedback that we get from our forums and convert it into Jira tickets, which are immediately assigned for work. That’s great, I really like that approach.
“Paradox is for people who like thoughtful games”
— Paradox recently released Victoria 3, which sold a whole 500,000 copies in the first few weeks. What do you attribute that to?
— Lots of players wanted to see what the sequel of Victoria 2 was like. There are players who love the grand strategy games (GSG) genre, and they try to play everything in it. For them, it was interesting to see how Victoria 3 was different from other grand strategy games.
— Did any difficulties arise during the localization process?
— Naturally. The example I gave about Turkish and the different letters “i” is a problem that arose during the localization of Victoria. We sorted out the Unicode and the support for those symbols, as well as the fonts—one of them was a custom font that had been developed specially for Victoria.
It’s also the first Paradox product that has support for Turkish from the outset. Victoria became fairly successful in Türkiye, and our colleagues said that the Turkish players were very happy.
— What’s your favorite Paradox project? Do you actually play the games that you work on?
— If we’re talking GSG, then my favorite project is Stellaris, because I was already following it when I worked at Wargaming. As a result, I started to play it and spent quite a lot of time on it.
I also really like Paradox’s approach to publishing other games that weren’t developed by us. We try to find games that fit the company’s culture. By that I meant that they’re not mass market, they’re reasonably niche products that require you to sit, think, and strategize. They’re projects for people who like thoughtful games.
I didn’t use to know that some games were released by Paradox. I played Mount & Blade for hundreds of hours, and only afterward realized, “Hey, it’s Paradox! Well, OK.” And there are lots of other really good products: Shadowrun, Pillars of Eternity, Tyranny.
— What recent games have impressed you?
— I used to only play on PC — I never had a console. For that reason, I missed a lot of exclusives that only came out on PS. Now that I’ve bought a PS5, I decided to play the first God of War because It’s an amazing game and I really enjoyed watching walkthroughs back in the day. Recently I’ve finished Ragnarok — a great sequel!
And in general, I’ve played so many hours in a huge number of games. My personal preference is for story-driven games, ones that have a really good plot. For me, that’s The Witcher, RDR, etc. Sometimes they’re open world, sometimes they’re linear, but the main thing is that they have a good story that really resonates.
I love “talking simulators” from Telltale, Don’t Nod, and Quantic Dreams and turn-based games, so I’m really looking forward to Baldur’s Gate 3 and the sequel to The Wolf Among Us. I’m also going to play our own Age of Wonders 4 and The Lamplighters League!