David J. Peterson, best known as the creator of of the fictional languages Dothraki and High Valyrian for HBO’s “Game of Thrones,” spoke at UR as a part of the River Campus Libraries’ Neilly Lecture Series. An old friend of UR Professor and fellow language creator Sarah Higley, Peterson discussed his approach to creating constructed languages, or “conlangs,” and the use of cultural, historical, and linguistic aspects in language invention.
Peterson received his bachelor’s degree in English and linguistics from the University of California, Berkeley, and his master’s in linguistics from the University of California, San Diego. In addition to his work on “Game of Thrones,” he is the author of “The Art of Language Invention” and of “Living Language Dothraki,” and has created languages for several other films and television series, including Trigedasleng, from the CW’s “The 100,” and Castithan and Irathient, for “Defiance.”
Mira Bodek: You attended UC Berkeley and majored in English and Linguistics. Can you talk more about your educational background and how you became interested in creating languages?
David J. Peterson: My intent in majoring in English was to eventually teach high school. That was the only thing I intended to do in going to college. I figured along the way I would take as many languages as I could, and Berkeley offered tons; in my first two years, I took Arabic, Russian, Esperanto, and French, and that’s how I discovered linguistics. My mother prompted me to study linguistics, but I didn’t think it would be something I enjoyed because it’s studying languages abstractly and I wanted to learn languages, so I thought it was pointless. But I took it anyways because it fulfilled the breadth requirement, so I figured, why not?
After my first class, I fell in love with linguistics because it was very easy. Our homework wasn’t reading novels or writing papers—it was a worksheet with data about a language you had never heard of and you had to put together the pieces, like a game. They aren’t random pieces—they’re sounds that fall together in a specific pattern. It was very easy, and it was a relief after going so long in my English major because it was so refreshing and so different. And it was in my first linguistics course that I came with the idea to create my own language. I thought, “Wouldn’t it be interesting if there was a language that did X or that did Y?”
MB: You said once that you begin creating your languages with an idea, usually a grammatical one. So did you use creating languages as a way to explore the ideas, theories, and concepts that we don’t already see in existing languages?
DJP: Well, it goes two ways.
One, it was helpful in learning things that I was introduced to in linguistics. Like if you learn the rules of a sport abstractly, like you read it on a Wikipedia page, you understand it in one sense. And then when you play the sport, you learn it in a different sense. So, in linguistics, it’s one thing to look at the data and the grammar and consider the challenges people who learn this language will run into, but it’s very helpful to get out of the English mindset. So, when you’re given foreign data, and you try to understand it in an English way, it’s helpful to explore it in a new language. So, that was very helpful for me.
Two, we always try to create something brand new. In the conlang community, there’s a phrase that says, “A natlang already does even weirder.” People will come up with things that they think are very novel and people will respond and say that there’s already a language, usually from Australia, that does that, and it does things that are more bizarre than that, and here’s the data.
MB: So, do you feel that conlang creators are constrained by the languages that they already know?
DJP: Well, no. But it is certainly true that the languages you already know will tell you things in languages that can work. So, we try to study as many languages as we can because there’s so much variety out there, and you need to know what’s out there to go beyond it, because you might create something that already exists.
MB: Which linguists have been particularly influential in your work? Have any conlang creators, such as J.R.R. Tolkien, influenced your created languages?
DJP: Starting at Berkeley, definitely John McWhorter—who studies Pidgin and Creole languages—really changed how I view languages completely. After that, definitely George Lakoff, whose work on conceptual metaphor was another layer in looking at languages. You know metaphor isn’t just a literary device—they’re very practical. You know saying you’re “in love” with someone, that’s a metaphor. Once I got to graduate school, the next layer was Farrell Ackerman, who presented a different way of looking at languages and, specifically, looking at morphology. And the last bit was Joan Bivi and her introduction to grammaticalization, and how everything in language evolved from a natural process.
MB: For “The 100,” you created an evolved version of English, and, in “Game of Thrones,” you created the corrupted version of High Valyrian that is spoken in Essos and in the Slaver’s Bay cities. How did you approach corrupting or evolving your own created languages?
DJP: The characters call it corrupted—in reality, it’s just evolved. The relationship between High Valyrian and the language spoken in Astapor is roughly the same between the one between Vulgar Latin and Old Spanish. Back then, when Spanish and French were first emerging, people would say that they were speaking Latin very poorly, not understanding that these were new languages that were emerging. And that’s what you see in “Game of Thrones.” First, I created the language that gave rise to High Valyrian in prehistory, then I created High Valyrian, and then I used High Valyrian to create the language of the Astapori. And it’s very realistic, how it’s discussed in the books—the characters say that it’s bastard Valyrian or bad Valyrian. And it’s also realistic that the upper class people maintain their fluency in High Valyrian while the lower class speak the evolved version. Both exist at the same time, the spoken language and the real language. Most of the time, the work I do is behind-the-scenes. I created the precursor to High Valyrian, but you’ll never hear it. Here you get to hear both.
MB: You talked about creating a hieroglyphic written version of High Valyrian. Is that still in the works?
DJP: I really want to do it, but I wouldn’t unless it became canon. So, I would need the approval of the HBO folks or of George R. R. Martin.
MB: Going back to what you discussed about metaphors: In the show, we hear that Dothraki doesn’t have the word for “thank you” or the word for “sea,” so there are these untranslatable words and phrases that don’t have one-to-one translations. What are your thoughts on things that are untranslatable, and, further, on translating specific aspects of languages like metaphors, subtext, or sarcasm in a conlang?
DJP: The story behind the “no word for thank you” thing came from me, but not in the way you think. In my original pitch for the Dothraki language, I wrote that Dothraki doesn’t have a word for “please,” which I stand behind. I think you can get away with not having a word for please, but I didn’t suggest that there was no word for thank you. I don’t think that’s possible. But then they saw that and liked that and wrote that line into the script. It didn’t originally exist. And I saw that and I was like, “Oh shit.” So, that’s that. Not very realistic and I’m not happy with it.
But it’s interesting when you’re creating a language for a work. A story is very linear in time. And so, it’s almost like you can imagine that, if you were in the “Game of Thrones” world, you could see everything at the same time, but in a story, you’re on a track taking you from to place to place with a limited view. We only see Dothraki from a perspective of an outsider, and we only see them in very unusual and stressful situations, things that happen once in a thousand years—the red comet and the dragons. So, we don’t see their day-to-day lives and what everyday life is like. But as a language creator, you’re creating the language for everyone.
And so, usually, when you create languages, you have all the nuts and bolts and then you present snippets, a story or poem or cultural description. So, you imagine a big world and then present small bits of it in a nonlinear, non-narrative way, so that you can actually see what language is like during dinner, or battle, or whatever version of court they have. So, you have different registers and different types and classes of people speaking. But you don’t see it in a greater narrative, though I’ve created it, but you won’t see it in the show. So, that’s my long answer where I’m saying “yes,” I have considered metaphors, registers, classes. It’s something we consider, but it never comes up.
Something I always wonder is what the Dothraki discussion is like when preparing their elaborate feasts—like a “Downton Abbey” version of Dothraki. All those small stories of people doing things and living their lives. But when you’re telling a story, you want to focus on the big parts, the dynamite, so you don’t see those things.
MB: Once, you said creating languages is very different from learning one, because when learning one you have to speak it with others and converse with them. So what makes a language challenging to learn? What makes it easy to learn? What are the ones that you’ve created that are easy or hard?
DJP: There was a really great infographic that I saw that showed how difficult a language is to learn based on the language that you speak. Because that’s really the question in asking what’s the most difficult language to learn, because they’re all about the same. But if you grew up speaking one language, other languages will be more difficult than others. It depends on how similar or different the grammar and vocabulary are. For example, it’s probably easier for an English speaker to learn French than German, but it’d be easier to learn German than Chinese, but also easier for an English speaker to learn Chinese to Japanese, because Chinese has some structural similarities; Japanese is almost backward from English.
For my created languages, Trigedasleng [from “The 100”] is probably the easiest. It’s probably the easiest language for a native English speaker to learn, period. Especially for younger speakers, older speakers might be very confused by it. Dothraki is not too bad. High Valyrian is a little more difficult because, grammatically, it’s a little more backwards. When it comes to English, Dothraki is a little more similar in that it’s head-initial. Castithan and Irathient [from “Defiance”] are more difficult to learn. Dothraki is probably second-easiest because there are a lot of similar structures that are familiar to English and because it’s lightly inflectional, but Irathient, which is grammatically similar but has lot of individual grammatical pieces that are alien to English. We’re used to Dothraki because it’s similar to French, Spanish, and German.
Trigedasleng is definitely the easiest. I remember people asking, “How can Clarke be fluent in Trigedasleng after only three months?” and I’m like, “How can you not be after two weeks?”
MB: In “Skyrim,” the Nordic language is a direct, literal translation from English. Do creators of conlangs judge other created languages?
DJP: It’s not even a language—it’s just a very stupid way of speaking English. The best ones come out of the community. Okuna by Matt Pearson is great. Or Sarah Higley, a professor here at the University of Rochester who I’m visiting and staying with—her language, Teonaht, is outstanding. Ithkuil—a very different type of language, by John Quijada—is also just one of the best that we’ve seen. The best created languages come from within the community, from those that have dedicated their lives to it, and I have a number of favorites.
MB: Of all the languages that you’ve studied, which is your favorite? Is it because of how it sounds or because of grammatical beauty? How has it influenced your conlangs?
DJP: Arabic was the first language that I studied, and the grammar absolutely blew me away as something very different and very elegant. I knew about the script beforehand and I liked that, but the grammar amazed me—I never knew language could work that way, and we haven’t seen anything else like it.
But Hawaiian is probably my favorite because I love Polynesian languages, and Hawaiian has the best phonology amongst all of them. I love Hawaiian. It’s not crazy exciting, especially if Arabic is in the discussion, and it’s not especially difficult because it’s head-initial, sort of like Dothraki. But there’s no inflection at all, no single or plural pairs—it’s all auxiliaries or particles, so it’s fun grammatically. It’s the most beautiful language I’ve ever heard. It’s funny because I love Arabic too, and there couldn’t be two more different-sounding languages. And I love both bits of it.
I try, when creating languages, to make them sound how I want, but I’m much more constrained than any language creator because the producers have to sign off on how it sounds. There’s this idea that if it has the sound “chuh” in it, it’s ugly, and if it doesn’t, it’s beautiful, and that makes me upset. They hear “chuh” and they think German, Russian, Arabic, even though the sound also exists in French and Spanish, and no one calls French ugly. It’s clear that what’s happening here is cultural stereotyping and ignorance about languages. But they’re also the ones that are in a position of authority, so it’s difficult to make my languages sound different when I’m constrained to draw from a small set of sounds that are deemed acceptable. And especially the other constraint that it has to be easy to pronounce for native English speakers.
MB: Do you ever feel constrained to what you can pronounce?
DJP: No, I’ll give myself challenges. It’s always funny to see what’s difficult to pronounce. Two of my most difficult-to-pronounce languages you wouldn’t guess. One is an early language I created before I worked on shows called Kamakawi, which started as a homage to Hawaiian. All the sounds are simple—very simple, very small phonology, sounds that every English speaker already knows—but the syllable structure makes it very hard to speak. It’s not like there are consonant clusters, it’s all consonant-vowel, but if I had a whole sentence, I fumble and screw it up. The other is Castithan, which I create for “Defiance,” which I tried to make easy to pronounce quickly, but it turns out I did the opposite in creating the sequences. So I had to do the lines again and again, whereas the theoretically more difficult Irathient language, which is phonologically more different, came out much more natural. I don’t know how the cast did it.
MB: Speaking of constraints from producers, did you feel constrained by the pre-existing words from Dothraki or Valyrian?
DJP: Oh, of course. And this is what a lot of people I work for don’t get. You create one word, one word only for a language, and you have constrained it beyond imagination. All those sounds are present now in the language. Because you’ve defined implicitly the syllable structure, the word structure, and most of the time it comes of out this well of what the English-speaking imagination imagines a foreign language sounds like. So you have the “th” sound show up, which, cross-linguistically, is very rare in all the world’s languages. If you create 17 languages, only one should have the “th” sound. And it’s in most of the languages I create for the show because it’s in the name of it—Dothraki, Castithan, Irathient. I’m not coming up with these names, I’m given them.
And also with Dothraki, if you look at the structure, it’s very reminiscent of Indoeuropean languages with its cases. You have the verb, with its minimal number of conjugations, and then you have your nouns, with their minimal number of declensions—so discrete units in chunks. I had to do that because the few phrases in the books were like that. You could’ve done something very different with the Dothraki language, even with the names that were just there defining the phonology. You could’ve done something like Inuktitut, with really super-long words with noun incorporation. You could’ve done more agglutinating, but still something more toward the European scale—like Swahili, with very long verbs and no nominal inflections.
But even if I could’ve justified it as a language creator, a clever trick for why the language in the book looked different from how it sounded in the show, it wouldn’t have been faithful to the character of the language. It would be really strange if normal Dothraki was these monster-long words and, in the book, there were these sentences with these short words. Even if I had an explanation for why that would be correct, it wouldn’t have been satisfying to the fans who were expecting Dothraki to look like how it was in the book. I wanted to be sure to satisfy the book fans because if the show was a flop and was one season and then cancelled, then I would’ve had a future with the book fans.
MB: Do you have any advice for aspiring linguists or conlang creators?
DJP: It’s similar advice to aspiring writers, like to read great stuff and write a lot and get good feedback, and it’s similar for creating languages. So study as many languages as you can, especially outside your language background. If you’re studying Spanish and you speak English, it’s great to study Italian and German, but it’s better to study Japanese and Mandarin, things that are very different, to get a sense of what the world is like, and to study other created languages to see what people have done. There’s a lot of great work online of people who just created it for the joy of it. See what’s been done and hasn’t been done, what’s been done well and what hasn’t been done well. And then do it. And if the project isn’t sustainable or worth the time, start a new one. That’s how you get better.