The Daunting Task Of Designing Language For Video Games
To feel a connection with the virtual world, we first have to believe in it. With sci-fi and fantasy settings in particular, designing fictional languages can help paint a more realistic world that captures the diversity and personality of its characters. Would Far Cry Primal’s prehistoric setting be conceivable in modern English? What if Skyrim’s ancient draconic tongue didn’t exist?
Whether they are pushing creative player engagement with ambiguous speech as we see in The Sims, or mimicking real-life dialects to give the impression of multiculturalism in Dragon Age: Inquisition, video games are heightening the bar for immersion. We chatted with various developers about how they approach this unique and daunting task.
This article was originally published in issue 279 of Game Informer Magazine.
Reviving The Past
When Ubisoft decided Far Cry Primal would be set in 10,000 B.C., the team faced a glaring problem: How would they coherently tell a narrative if all the characters are primitive cavemen? Would groans and grunts suffice? Before settling on Proto-Indo-European, Ubisoft first explored other routes, such as tasking its writing team with creating a fictional language from scratch. After realizing it would be too much of a strain on the writers, they looked for outside resources.
A member of the localization team suggested two linguists, Andrew and Brenna Byrd from the University of Kentucky. They had prior expertise in researching and translating Proto-Indo-European (PIE), a rudimentary language that was spoken between roughly 2500 and 4500 B.C.
“Toward the beginning of the project, when [Ubisoft] sent us an email, we actually thought it was spam,” Andrew Byrd says, laughing.
The Byrds translated the game’s 40,000-word script, which was previously written in simplified English. The colossal project tasked the two with building a fictitious language based on variations of PIE for three of the game’s tribes: the Wenja, the Izila, and the Udam. Each tribe’s name is inspired from a combination of words that best represent the tribe’s mindset. For example, Wenja is derived from the word “wan,” which means “to hunt.” The Izila speak a language of their own — a more melodic iteration of Wenja. While the Wenja and Udam can understand each other, their dialects differ phonetically in a similar way that Southern English and New York English would.
Like any complete language, Wenja has a consistent grammar and syntax. Sentences follow the following structure: subject + object + verb. While it’s possible to conjugate verbs in Wenja, it is not possible to differentiate between past, present, and future tenses. By enlisting the help of friends from grad school, the Byrds turned the project into a large, collaborative activity, where they would cross-analyze each other’s work and mentor one another.
Ubisoft wanted the Byrds to base the language off a more archaic version of PIE, which the linguists call Proto-Proto-Indo-European. Modifications were made to better suit the setting, and to accommodate voice actors who struggled with specific words. While the Byrds estimate that 95 percent of language in Far Cry Primal are reconstructions molded from PIE, the remaining five percent are educated guesses.
In PIE, there are nearly 20 different roots for the verb “to shine,” but modern words including expressions as simple as “yes” don’t have an existing term. The two had to improvise with creative solutions, such as “yes” becoming “it is correct.”
“[Accuracy] was something that was very important to all of us in the project, because Proto-Indo-European is something that many, many people have studied,” Brenna says. “We wanted it to be as authentic as possible. We wanted it to be believable for other linguists.”
The result was a rich, complex language that breathed life into the land of Oros and gave an aura of realism. A problem the development team soon realized, however, was that the foreign language would mean a heavy usage of subtitles. Ubisoft didn’t want players to spend their entire time reading, so they put more emphasis on motion capture to drive home the narrative through body language and facial expressions. To test this out, the team revisited their previous entry, Far Cry 4, but this time entirely in Japanese.
“We wanted to see if we were able to understand what was going on in the world without understanding the language,” says narrative director Jean-Sebastien Decant. “And of course, the body language, the emotion, and the situation made it very clear that people would get what was going on.”
Next: Learn how the team at Maxis formed Simlish, The Sims' fictitious language.
Transforming Speech From Gibberish
Far Cry Primal’s language-building posed many questions. Could the game remain immersive even if players didn't understand a word? Would their interest wane? The Far Cry 4 test affirmed to the team that a story told entirely in an unknown language was conceivable if done right. In comparison, The Sims’ creators at Maxis built a language that was supposed to be unintelligible, so players could create emergent narratives of their own.
During the development of the original Sims, which released in 2000, creator Will Wright and his team experimented with foreign languages such as Navajo for dialogue, but that idea soon fell through after the voice actors struggled with the performance.
“We brought in some improv actors from the Bay area, and we gave them those languages, but they were very confusing for them to work with,” says audio director Robi Kauker. “They didn’t know where to put their emphasis.”
Two of the original Sims actors, Stephen Kearin and Gerri Lawlor, brought a new idea to the drawing board. They informed the team of an improv exercise where the goal was to build a coherent story using only nonsensical sounds such as hums and mumbles. From there, early iterations of Simlish were born. After seven days of recording improvised sessions of Simlish, Kauker and fellow audio director Kent Jolly spent six to eight months editing pieces of audio, mixing them up in new combinations to build the Simlish we hear in the game.
As the series matured, so did Simlish. A basic vocabulary began to form and phrases became more consistent. For example, “Sul Sul” is known as the hello greeting and “nooboo” means infant. The language is simple and much of it remains gibberish to this day, except for a short list of words that have their own definitions. While some vocabulary defines certain objects, many of the words have multiple meanings relating to various emotional inputs. This was especially prevalent as Maxis moved forward with The Sims 4, which puts emphasis on emotions more so than any other Sims game.
“We have a lexicon, but it’s not a one-to-one translation by any means,” explains The Sims 4 audio director Jackie Gratz. “It’s more [based on] emotional context. We have five or six other words that in general express that you’re enjoying eating something, or something is really grossing you out, or something has a flavor to it. But the words can be used in many different contexts within that general idea.”
The team kept much of Simlish ambiguous for players, leaving it up to their own interpretations. Considering the series is user-generated in many ways, from home building to in-depth character creation, it comes as no surprise that the language, too, was there to breed creativity.
Advantages Of Fragmented Language
A full-fledged fictional language is an ambitious task, but it’s not always necessary, as we saw with The Sims. At BioWare, creating condensed, fragmented language was more efficient for its fantasy role-playing franchise Dragon Age, due to the large size of the writing team and the extensive lore of the game. Simulating a sense of realism was a primary goal, but sacrifices were made so intricate details wouldn’t end up lost in translation. They settled on a compromise: The team designed bits of language and borrowed inspiration from real languages to build a world that is as believable as it is multicultural.
With an internal language database of grammar rules and vocabulary, any writer in the team can glance at a list of words quickly to grasp the many fictional languages from Dragon Age.
Thedas, Dragon Age’s sprawling fantasy setting, is a continent filled with several empires and kingdoms home to elves, dwarves, humans, and others. While English (usually called “the common tongue” in-game) is spoken universally, each area comes with its own dialect, and several races have their own language they revert to occasionally with brief phrases. Most dialects and accents resemble real languages, such as the Orlesians speaking with French accents. The team saw this as the best compromise. The latest entry to the series, Dragon Age: Inquisition, covered more ground than any other installment. The deeper the lore and cultures became, the greater the team wished to flesh out fictional languages too.
For the languages that are completely fictitious, it’s a different story. The two most complex languages, and most heard in the series, are Qunlat (spoken by the Qunari) and Elven (spoken primarily by the Dalish elves), though these are far from fully realized languages. BioWare describes them as being closer to ciphers.
“In both cases, they are languages that are good for communicating simple feelings, or pointing at something and saying what that thing is,” says BioWare writer Patrick Weekes. “They’re not languages that you can use to get instructions on how to program your microwave.”
This was done purposely, since a language with extensive vocabulary and grammar would mean each writer would have to learn a new language from scratch. But the purpose of this direction goes deeper: Fragmented language is used to illustrate how little is known about Elven in the world.
“It was deliberate that it was just fragments, because until you get to Inquisition, there’s not a single elf in any of the Dragon Age games who knows how to speak Elven,” says Dragon Age writer Mary Kirby. “The language is kind of a lost part of their culture, so they have phrases and words and that’s it.”
Qunlat features more vocabulary than Elven, though once again it remains fragmented. The Qunari are a brutish race that uses oral communication less frequently than other races do. Their speech is stripped to mere necessities, and most words have several meanings.
“Qunlat is harder for the writers to write, because it’s very metaphorical,” Kirby says. “Elven is the hardest for the actors, because it’s a little bit sibilant, and it turns into a giant tongue twister. I’m very careful that every time I include a message of long Elven in dialogue, I actually add an apology in the VO notes.”
Dragon Age’s languages have very little conjugation with few grammar rules, and aren’t meant to be dense enough for fans to speak to one another in. Instead, more focus was put on intonation and phonetics. Elven, for example, is spoken without closing your mouth often because of its scarce usage of hard consonants, such as the phrase “Dareth Shiral,” meaning safe travels. Teaching voice actors the correct pronunciations is a challenge of its own.
Next: Find out how Far Cry Primal's cast learned a new language from scratch, and how Skyrim built its own runic alphabet.
Teaching The Cast A New Language
It’s common for voice actors to first struggle with learning a new language. With Far Cry Primal’s PIE-inspired language being exceptionally in-depth, it required teaching and mentorship from the linguists. Every morning, Andrew Byrd would hop on Skype with voice actor Elias Toufexis, who plays the protagonist Takkar, to teach him the proper annunciations of Wenja.
“We would get together every morning on Skype and practice his hundreds of lines,” Andrew Byrd says. “He’s such a talented actor that he needed very little coaching.”
The rest of the cast, particularly those that took part in motion capture, were guided through the language by Brenna. She taught three immersion courses on set. The actors would introduce themselves in Wenja, and speak it for the entirety of the hour. Then Brenna shouted out commands in Wenja, such as telling the cast to run around the room or pretend to throw objects. This is a technique known as Total Physical Response, an instructional tool that has been used in language classes since the 1970s.
“The actors would not speak, only react physically to what I was communicating verbally,” Brenna says. "It’s a little like the game Simon Says, except the challenging part is remembering what the words mean.” It became a chaotic but fun activity that the two linguists believe was a turning point in the project, allowing the cast to become more comfortable with the language.
“As you can imagine, it’s really hard to learn a completely made-up language,” Andrew Byrd explains. “Not only do that, but coordinate it with new movements that are completely foreign to you, and express emotions within the cinematics – it was a very, very difficult thing for the actors. But in doing these lessons, Brenna really created a community among us.”
Creating An Alphabet
Building a believable digital world is a difficult task. The Elder Scrolls series has long had a rich world full of elaborate detail, and its dragon language, with over 60 individual words that combine to form 20 dragon shouts, reinforces this ideology. Largely created by senior designer Emil Pagliarulo, every word in the dragon language is translatable. Even the iconic dragon shout “Fus Ro Dah!” can be broken down into English: “Fus” means force, “Ro” means balance, and “Dah" means push.
When we spoke to Pagliarulo for our 2011 The Elder Scrolls V: Skyrim cover story, he explained the many challenges of writing the stanzas for “The Song of the Dragonborn,” a tune that would capture and sway the hearts of many fans. He had to make sure it would rhyme in both English and the dragon language, while also communicating an epic tale.
The language itself is comparable to BioWare’s use of fragmented speech in Dragon Age. Rather than designing a complete language, the dragon language abandons most grammar rules and syntax, and is instead considered word-based and hieroglyph-based. However, Bethesda went one step further, by building an alphabet that would be scribbled across Skyrim’s many ancient ruins and word walls that teach you new dragon shouts.
Pagliarulo worked alongside concept artist Adam Adamowicz to create the runic alphabet, which is comprised of rough scratch lines to imitate the talons of a dragon, as if they were engraved by the large beasts themselves. Adamowicz created a font with those aesthetics in mind, and designed symbols for different letters, though not all of them correspond to the English alphabet.
In the dragon language, there are 34 unique runic characters, and certain Roman alphabet letters don’t exist, such as “c.” Other times, a single symbol can represent several Roman letters at once, such as double vowels, including “ii” or “ei.”
Building Believable Worlds
Whether it's using internal wikis for consistency or hiring linguists to build entire languages, developers are paying more attention to what their language conveys to the player about the universe they're stepping into, and as we've seen with these various games, it's paying off. Creating languages is an ambitious task, but the value is substantial: Boundaries continue to be pushed to portray a living and breathing world, one word at a time.