Visit theallusionist.org/verisimilitude to hear this episode and find out more about it.
This the Allusionist, in which I, Helen Zaltzman, swing into the languageverse.
Coming up in today’s show, we find out what it takes to make a fictional language sound realistic, even when the speaker is riding on a dragon.
On with the show.
DAVID PETERSON: I became very interested in learning languages, just as many as I could. My last year of high school I became really interested in language, and so when I got to Berkeley for college I wanted to study as many languages as I could. And so I started taking Arabic. I took Russian and one of the languages that was offered - it was advertised actually on the door of my dorm - was Esperanto. It was a student-taught course,and I'd never heard of anybody creating a language before. I thought to myself, "What if I created a language of my own?"
HZ: So, he did.
DAVID PETERSON: I found it immensely rewarding and just like an extremely addictive. And so I just kept up with it for the next 18 years, 18, 19 years I guess.
My name is David Peterson and I'm a language creator and writer. I am best known for being the creator of the Dothraki and Valyrian languages on HBO's Game of Thrones.
HZ: Approximately how many languages have you invented at this point?
DAVID PETERSON: I think I've invented over 50 languages at this point. Not all of them are very large in terms of vocabulary size, and not all of them are very good. I had created about 17 before I ever started working on Game of Thrones.
HZ: The languages you hear in Game of Thrones: Dothraki -
[CLIP] Khal Drogo: “Moon of my life, are you hurt?”
HZ - the various dialects of Valyrian:
CLIP: Daenerys: “Valyrian is my mother tongue.”
HZ: - those aren’t the actors making up some gibberish. Those are functional languages, with large vocabularies and complex grammars and etymologies.
HZ: And why was it important to have a functional language?
DAVID PETERSON: Important to whom, is the question. So for the producers, they had actually tried their hand at creating some gibberish for the auditions, and they were just really really disappointed with what was happening in the auditions. The actors were good, but it just sounded silly; and so they decided they should just hire somebody to do it now. I suppose technically you could get by with just, "Here comes the line and here's some Dothraki-sounding stuff" but it's going to sound too fake. It would be too hard to actually make it sound good. And people would wonder, especially for elements where they focus on a particular word, why the same word isn’t appearing in a line later on in the series where that word should be.
HZ: Before HBO started shooting the first season of Game of Thrones in 2010, the show’s producers approached the Language Creation Society - cofounded by David Peterson - and they held a competition to find someone to construct Dothraki language for the show.
DAVID PETERSON: The person who won this competition was going to be representing the language creators, period. And so there was no question that person is going to have to do the best they absolutely could. And that's what I tried to do at that time.
HZ: There weren’t too many parameters for the language creators: they had to translate the script for the pilot, and provide a phonological outline for their version of the language, as it was going to have to be possible for the actors to say it as naturally as if it was natural.
DAVID PETERSON: But with no cap on the amount of material that you could provide, there was gonna be somebody that was going to go way overboard and produce way more material than was just humanly reasonable. And so I felt like I should be the one that did that. Otherwise I would always be wondering in the back of my head, “What if I had done a little bit more?”
HZ: How much did you come up with?
DAVID PETERSON: For the first round of the competition, I had over one hundred and eighty pages of material; that was a couple of weeks’ worth of work. And then I got to the second round and realized I would have to do more. So for the second round it was beefed up to just over 300 pages of material. Another couple of weeks.
HZ: In George R.R. Martin’s Song of Ice and Fire book series that the show is based on, there were only 56 Dothraki words, 24 of which were proper names. David decided he was going to honour the language as it was in the books, as, though there was little of it, it did work.
DAVID PETERSON: It did. It did; everything he did worked. Everything he did worked perfectly. And not only that, it wasn't just copied from English. That was wonderful. I think he has a really really good language sense. While it coheres, and it's completely 100 percent consistent, it's not especially interesting in the sense that it's very much like reminiscent of a romance language. So in other words, if the only idea you had of how a language might be foreign was something like Spanish and French and you didn't want to create something was exactly like Spanish and French but you want to create something foreign-ish, that's basically what Dothraki is, in terms of what George R.R. Martin has got in the books.
HZ: David built out the language for the nomadic warrior Dothraki people. It had no written form, it had no word for ‘please’ and no word for ‘toilet’.
DAVID PETERSON: They don't have a word for toilet because toilets don't exist.
HZ: He won the contest, and when you’re watching Game of Thrones and hear Dothraki or High Valyrian or its informal register Low Valyrian, that’s all David’s work. He’s since created languages for films including Thor: The Dark World and Doctor Strange, and TV shows including Defiance, The 100, and Penny Dreadful.
DAVID PETERSON: Most of the languages I'm called on to create now are naturalistic languages, that is they're supposed to be for realistic human beings. It requires a lot of work to actually produce something that qualifies as naturalistic. When it comes to creating a naturalistic language, if you're going to evolve it or if you're going to simulate the evolution of a natural language, there are three elements: the three elements are the sound system of the language, that its sounds change over time. The second element is the meanings of words. So meanings of words will change over time. And then the third is the evolution of grammar, that is grammar itself actually evolves out of something over time. Basically my entire goal is I just make sure that all of the parts are realistic and naturalistic. That is a realistic sound system, a realistic inflectional system, a realistic grammar and a realistic distribution of phonemes when it comes to the lexicon. Then the result will be naturalistic.
HZ: Want to create your own language? Here’s what you should bear in mind:
DAVID PETERSON: I guess the first question for any conlanger is: is this language a priori or a postieri?
HZ: An a postieri constructed language is one based on natural languages, like Esperanto draws on several Indo-European languages. Whereas an a priori constructed language has unique vocabulary and grammar.
DAVID PETERSON: That doesn't mean that they will never have existed before especially with grammatical properties, because there are only so many ways that languages can differ; but it will mean that they were not actively borrowed from another language.
HZ: Don’t just translate your natural language into your new language. So many of the ways we use words are idiosyncratic, allusive, reflecting the societies we live in: a literal translation isn’t going to work.
DAVID PETERSON: So this is something that really cannot be avoided. The question when it comes to creating naturalistic language is: are you simply copying the metaphors of your own language, or are you actively engaging with the notion of conceptual metaphor and crafting different and unique but nevertheless realistic conceptual metaphors? That is, if your speakers are human, there will be a lot of the same metaphors; some of them will be different. And so yeah, when you're creating something like that, you should be actively engaged in how that works. So that you can be crafting something that is realistic and is not simply copied from your own language.
HZ: And if you’re creating a language for a fictional world, don’t forget to consider the etymology. The people and societies for which you’re creating the language - how did they come to be the way they are at the time they’re being depicted?
DAVID PETERSON: I try to go back a thousand or so years. Just imagine a thousand or so years - that's usually a good time depth. Before that, things just get dark and hazy.
HZ: Yeah, there's not a lot of written record.
DAVID PETERSON: So usually that's going back far enough. And I try to go back far enough to imagine the culture in a place where there wasn't a lot of movement, where it was more insular, and they were able to create words or come upon words that just reflected their original culture and original environment before they started branching out a lot. Then I just carry it forward from there and imagine who did they meet, when did they meet and how did that affect the culture in the language.
HZ: And when you're creating multiple languages for a world, like in Game of Thrones, do you act like they have a common root like their own Proto-Indo-European? Do they have overlapping vocabulary or loanwords or anything?
DAVID PETERSON: They have a common root if they have a common root. If they don't then they don't.
HZ: Then, if your language has a written form, you’ve got to figure out what it’s going to look like.
DAVID PETERSON: That's what I think is the most fun, coming up with writing systems. But the nice thing is, there's a lot more math to it than you would think, at least in coming up with a system. If you look at English, just a standard font like Helvetica, and you break down the characters, you see that there's a very very small number of pieces that are repeated over and over again.
HZ: An upright line and a curve appear as a lower case b, d, p and q - the same shapes, flipped around.
DAVID PETERSON: If you think about it this way, it's actually not too bad to create a new writing system. You create these small pieces and then you put them together into larger arrangements, larger glyphs; and then flip it horizontally, flip it vertically, see what works. You need to have a sound system to guide you so that you know what you're doing; but it can actually be really simple to create something aesthetically pleasing. Because part of what we find aesthetically pleasing is this repetition and unique arrangements of familiar pieces, and that's something that's within anybody's grasp, because some of these pieces are as simple as just a line.
HZ: And think about what materials the people would have used to write the language.
DAVID PETERSON: You just imagine for the history of these people: what stages did they go through? What mediums did they use to write on and and what did they use to write with? When did it change? What other groups did they encounter? And then how did the writing system change in each of those stages? And then if you want to go full bore with it, you can actually get the real stuff and try it out. You can get stone and try to chisel into it. You can get wet clay. You can emulate palm paper with tissue paper and so on. You can get different types of styluses - you can make them if you want - and you can actually see what would happen. And in so doing you can take actual pictographs, which can be as simple as you want and turn them into characters at the other end.
HZ: So David here thinks about every part of speech, and how they relate to the other parts of speech, and phonetics, and the appearance, and the etymology of the last thousand years. He is thorough. He gets into language at a molecular level. But what’s the most difficult part of language to create?
DAVID PETERSON: Creating verbal systems is also the most difficult part, because they're constantly in flux. Verbal systems change for my generation and generation - they change even within generations, ever so slightly, and they have tiny little wrinkles that are just a nightmare to fabricate. And then of course they're ephemeral. So you go through the whole process of saying "Alright, this is how this is exactly going to work." But 30, 40 years later, people may be speaking differently.
HZ: And how big is the lexicon that you've created for Valyrian and for Dothraki?
DAVID PETERSON: I think Dothraki is just over 4000 words and Valyrian is just over two thousand. Lexicon size is just a function of time. I'll get there.
HZ: Do you invent swear words?
DAVID PETERSON: Yeah, I mean, that's usually the first thing everybody asks for, when I'm hired for this thing. So you create swear words all the time.
HZ: Given that you’re creating the language, you can iron out the problems you see in a natural language: yours doesn’t have to be imbued with a sexist or racist history. It can be perfect. Or can it?
DAVID PETERSON: If you're setting out to create a language that's more perfect than a natural language: that defies the very tenet of naturalism. So really the goal is just like when you go to Disneyland and they're trying to make something look like a mountain - provided it's in one of the realistic areas of Disneyland, not Toon Town or something, they want to make it look like a real mountain, they want to make it look like a cartoony mountain or a fake mountain, they want it to look like a real mountain. So if you make something that's more perfect than a real mountain where it's like "real mountains having all these crags and nooks and crevices, it would be more sensible if they were perfectly straight like we imagined the pyramids being,” and you created that, well, then you will have failed to create a naturalistic, realistic looking mountain. You have created something different. So if you're creating a naturalistic language, the goal 100 percent is verisimilitude; you have to make it look as much like the real thing as you can. And that means that it's going to necessarily have irregularities as a part of it, or just quirks and bizarre oddities, and fall within the realm of natural human language variation.
HZ: And remember, natural languages aren’t formed by one person; they’re formed by every user, over a very long time. Hence they’re riddled with irregularities and contradictions and eccentricities and mistakes that stuck. So add ‘realistic foibles’ to your conlang to-do list. This all may seem like a daunting amount of detail to have to think about, but don’t let it put you off trying to create your own languages.
DAVID PETERSON: If you want to create a language first just start because the only way to really get better is to do it. So you just got start. You just gotta get in there. Don't worry about like "Oh where should I start?" or anything. Just start somewhere, it doesn't matter. Just get your hands dirty.
HZ: Do you remember how you felt the first time you heard one of your languages in use?
DAVID PETERSON: The first time that I heard anything spoken - it was Dothraki - I immediately thought that the actor had made a mistake. It really bothered me. It turns out that I just misremembered how I translated it. And then I decided that I had translated wrong. After I heard it, I changed my mind about how I should translate it - nothing you could do about it at that point. But I think that set the tone. Generally I'm always kind of disappointed. It's never exactly how I imagined. Sometimes it's pretty good. But then every so often it's just really really good, and it’s very pleasing.
HZ: And are you happy with how it turned out?
DAVID PETERSON: Some of it. Yeah, some of it. I'm 90 percent happy with Dothraki. There are things that I avoided doing because I thought people might actually want to learn language. There are a couple of things I pulled back a little bit, made more regular, that I might not have otherwise. I thought that Dothraki was gonna be huge like Klingon; that never really happened. So I shouldn't have been super concerned with that. There are also things that I didn't do because of production; like having just a regular old word for ‘horse’, I might not have done that. I did that because you can never know what type of horse you're going to get on set. And so if they referred to it, I wanted to have a word that worked no matter what type of horse was there. But had I done it the way I wanted to, it would have been only words for specific types of horses, that's it; no word that just works for any type of horse. I think that would have been nice. And then also there was an element of the grammar, a couple of elements of the grammar, that I incorporated specifically because I feared getting fired and replaced by a staff writer after one season. And after I produced this big old huge language, I thought that they'd say, "Well now that we have this, we can just use it and we don't need you anymore."So I created elements of the grammar that are unstable, and they were intentionally ambiguous, so that no matter what translation was provided on the show I could always say that was incorrect.
HZ: That's ingenious.
DAVID PETERSON: Yeah, but at the same time it's not the best result; it's frankly a little confusing.
HZ: Who decides how it's pronounced?
DAVID PETERSON: I did. I looked at the material in the books and I decided basically, all right: this was created by an American English speaker. But it was created by an American English speaker who is trying to create something foreign-sounding. And so I had varying constraints. 1. What did I think he intended? 2. What do I think most readers who are American English speakers, what do I think they will assume when they read this? And then how can I balance that with my own linguistic intuition as to what makes sense. And then I also wanted to make sure that if something was spelled differently it was pronounced differently. And that worked pretty well for everything, except that word 'khaleesi', which is just a real thorn in my side.
HZ: Why is that?
DAVID PETERSON: Because there's no way it should be pronounced 'khaleesi' based on the spelling. So I had to decide: am I going to respell this thing because I know how people are going to pronounce this, or am I going to honour that spelling and then pronounce it differently? I made the latter decision, and I think that was the wrong decision. I think I should've changed the spelling. Changing the spelling of khaleesi to 'khalisi' would have fixed all of my problems.
HZ: The khaleesi is the wife of the ‘khal’, a Dothraki warlord. In the books, khaleesi is spelled K H A L E E S I.
DAVID PETERSON: You have an 'ee' and then an 'i'. They can't be pronounced the same, because they're spelled differently, if I want to honour George R.R. Martin's spellings, that is if something is spelled differently it's supposed to be pronounced differently.
HZ: So the double E in the middle should not sound the same as the I on the end.
DAVID PETERSON: So I decided to honour that, which gives us the pronunciation 'khal-eh-eh-si'. But I should have realized that there was no way I was going to win that fight with the producers, because the instinct is so strong with English speakers, when you see a double E it's pronounced 'ee'.
HZ: It’s not a problem that the English-speaking characters in the show mispronounce the word; it’s a foreign word to them, they might well get that wrong. The problem is this particular mispronunciation, pronouncing that double E ‘khaleesi’ like how it’s spelled, because they never would have seen it spelled - in their world, Dothraki is not a written language. And if they’d just heard it, they wouldn’t be pronouncing it ‘kaleesi’.
DAVID PETERSON: So the issue is that if the native Dothraki pronunciation is 'khal-eh-eh-si', then an English speaker who hears that is going to pronounce it either 'HA-lay-see' or 'HA-la-see' or 'ha-LA-see', not 'ha-LEE-si'. It would be kind of like if English speakers heard 'fiance' from French and pronounced it 'fian-cie'. That's just not going to happen.
And so I should have just respelled it on my own and just said that the word is pronounced 'khalisi'. I also could have rescued the stress, even if it were pronounced that way it should be stressed 'KHAlisi'. However, with the etymology that I provided, I could have just said that the stress switched to the affix, because that little affix comes from an old word that means 'woman' and it's in modifier position, it comes after the noun. So it's like 'female khal' is what it is. And I could have said 'khalisi'. Where if it was actually the older nominal compound khaleesi...
HZ: Stay with us, because I want you understand just how much thought goes into this constructed language…
DAVID PETERSON: ...there would have been a little bit of a bigger stress on the second word. And then I could have had it just be perfect. The Dothraki would have been 'khalisi', and all of them would have been saying 'kalisi', and that would've been fine. I should have done that, I didn't do that, and it's undoable now.
HZ: So the word ‘khaleesi’ is an impossible word, and now every time you watch Game of Thrones and hear it, or you meet one of the several hundred people who’ve been named ‘Khaleesi’ since the show began, spare a thought for David Peterson’s pain.
But let’s focus instead on what makes this worthwhile, constructing languages for fiction.
DAVID PETERSON: Well let me put it this way. If you're going to have a fictional world, first of all you should always be thinking about what the linguistic reality of that world is. It's something that gets ignored. This goes for whether it's a book or whether it's a show. Even with a book, even if you're going to write everything in English - which is fine, we expect that from books - you should be thinking about what is the linguistic history of this world. This is an entire world where everybody speaks the same language? But they also say like ‘ye’ and ‘thou’ every so often? Which, why? Or the other thing that you see often is everybody speaks the same language except this small group of people who speak another language, and they're usually feral or wild or bad or something. And it just makes the world seem a little fake, which, if you're going to the whole point of inventing a world, probably do something with it. Authors don't have big budgets, I get that. Movies do. TV shows do - certainly bigger budgets than an author, especially a first time author. And if you're going to spend any money on set design to try to make the sets look authentic, then spend some money on a language creator: you can afford it.
HZ: Do you ever think, “I'm putting all this work and I came up with four thousand words and a thousand years of etymology and most people are just going to read the subtitles and not really take in the spoken language as it's happening”?
DAVID PETERSON: Well that's fine. Look, there's always gonna be people that don't care, and that's fine. But for the people that do care: that's who you're doing this for.
HZ: David Peterson is a language creator - you can see his work on Game of Thrones and The 100, both of which return with new seasons in April 2019. And he’s just finished working on the upcoming adaptation of Dune. You can also read his book, The Art of Language Invention, for more detail about what to think about when constructing a language - and even if you’re not planning to construct a language, but just want to understand more about what a language IS and how it functions; his book went deeper into that than any other I’ve ever read. And in today’s Minillusionist, David talks about something I’ve been waiting to come up for the entire four-plus years of this show.
MINILLUSIONIST
HZ: Ever wondered why the plural of ‘mouse’ is ‘mice’ not ‘mouses’? Prepare to never wonder again.
DAVID PETERSON: The reason that we have that is because it was regular at one point in time: the word for mouse used to be 'mus' with a long sound and the plural used to be 'musi' where it had a little ee on the end.
HZ: I love that we're getting into i-mutation there.
DAVID PETERSON: Yeah yeah yeah. So you know where this is going! So what happened was this ee on the end, this ee sound, it affected the previous vowel. The way that it works was you could say 'musi', but if you say it a lot, what may happen is that you anticipate the place of the final vowel that you're getting to. And so what happens is you might move your tongue body forward to make it easier to say the next vowel.
HZ: Our lazy mouths are apt to shorten the distance between one move and the next. Try saying the word ‘doh ing’ - say it with me, ‘doh ing’ - that’s much more effort than ‘doing’, isn’t it? You can feel how much more you have to move to say an O sound then an i sound - say it with me, O I, now say oo I, oo I, see, now your mouth’s in eco mode, right? Aren’t we having some 7th century vowel shift fun? OK, back then, the plural of the noun ‘man’ was something like ‘manniz’, or ‘foot’ was something like ‘footiz’, goose ‘goosiz’. Put that in your mouth: goosiz. Goosiz. Say it enough times in quick succession and you find the word drifting upwards and forwards in your mouth, that first syllable altering to anticipate the second, menniz, feetiz, geesiz. And then English largely ceased being an inflected language, so dropped the suffix ‘iz’ from its plurals, so we were just left with that first syllable. Men. Feet. Geese.
DAVID PETERSON: So that is how a sound change changed the grammar of the language. But that's only half of the story. The second half is: why do we still do it? Because we have mouse and mice; we have louse and lice; we don't have house and hice, right? The reason that this one stuck around is because people interact - or at least for a while interacted - with the plural version of ‘mouse’ fairly commonly. So in other words, if you imagine going way way back, there was a lot more time people would encounter scurrying herds of little mice, as opposed to having, say, a pet mouse at their house. Same thing with lice. It's rare that you would have a single louse that you would encounter as opposed to lice. Consequently, people kept using the irregular plural of mouse quite a bit and so they didn't forget it. On the other hand, irregulars, whether they be irregular past tense forms or irregular plurals in English, irregulars that don't get use a lot, people will forget them and so they will become regular. So when you're creating these irregular forms in your own language, you have to keep in mind: alright, which of these irregulars are likely to stick around and which ones are likely to be analogized? But as a language creator, this is something that you keep in mind as you're trying to build something that is really kind of authentic and interesting and naturalistic.
The Allusionist is a proud member of Radiotopia from PRX, a collective of the most excellent podcasts on the interwaves. And as well as all of us long-running shows, your regular companions, we have a few short special series that you can hear if you subscribe to Radiotopia Showcase on your podgatherer of choice. If you haven’t already, do so now - NOW! - to listen to the new four-parter Spacebridge, a pretty bananas true story about the Cold War involving New Agers, technologists and astronauts, it’s like psychics vs the Iron Curtain. And it sounds amazing too, so put on headphones and prepare to be transported. Find Radiotopia Showcase, and all the Tope shows, at radiotopia.fm.
The all new Allusionist live show is coming to New Zealand and Australia in April, May and June. So if you’re in one of those countries, come to see it. I’m still finalising all the dates, but a couple are on sale already - get your tickets, Auckland and Sydney! The show is coming to a lot more places, so keep an eye on theallusionist.org/events because I’ll add new fixtures as soon as they’re on sale.
Your randomly selected word from the dictionary today is…
cancrine, adjective: crab-wise; (of verses etc) reading both ways, palindromic.
Try using it in an email today.
This episode was produced by me, Helen Zaltzman. The music is by Martin Austwick; hear his new album at palebirdmusic.com. Thanks to Nate DiMeo of The Memory Palace; for more constructed language fun, listen to the Toki Pona episode of the Allusionist, in which Nate and I attempt to learn a minimalist language and it breaks my brain.
You can find that episode, and all the other episodes, and all the transcripts and randomly selected words from the dictionary, and additional information about each topics, and upcoming Allusionist live shows, and links to the show on Facebook and Twitter, at the show’s forever home theallusionist.org.
