Creating Dothraki: An Interview with David J. Peterson and Sai Emrys |

Creating Dothraki: An Interview with David J. Peterson and Sai Emrys

Jason Momoa will protray Dothraki leader Khal Drogo

Last Monday, HBO issued a press release announcing that they’d hired David J. Peterson through the Language Creation Society to develop the Dothraki language for the upcoming Game of Thrones TV show. Both David and Sai Emrys, the president of the Language Creation Society, were kind enough to answer some of my questions about the genesis of Dothraki.

Fans of the books will remember Dothraki as the language of Daenerys Targaryen’s husband Khal Drogo and the rest of his nation (and by the end of the first book, Dany speaks it fluently, too). Dany is one of the main viewpoint characters of the series, but spends much of the series (so far) in exile, far away from her native country and the center of the action. Though a few words do appear in Dothraki in the books, for the most part its speakers’ words are translated into English. Author George R. R. Martin noted in his own blog post on the subject that despite writing in the tradition of Tolkien in other ways, he’s terrible at foreign languages and never made up more than a few words for any of the languages spoken in the series.

Constructed languages—or “conlangs”—are hardly an unfamiliar subject to fans of speculative fiction, Tolkien’s Elvish languages being a frequently-cited example. This is also not the first time an outside expert has been hired to flesh out the linguistic world of a fictional show: linguist Marc Okrand created the notoriously difficult-to-learn Klingon for the original Star Trek (and later the Atlantean language for the Disney film Atlantis).

The Language Creation Society’s website now hosts a page devoted to Dothraki that will be updated with more details as work on the series progresses. Fans are being encouraged to become active in discussions about the language, but according to this post, we’re not likely to see much information until closer to the premiere. However, at the end of this interview you will find an exclusive new bit of Dothraki to help tide you over.

Ellen B. Wright: Sai, what does the Language Creation Society do?

Sai Emrys: Our goal is to promote conlangs and conlanging through offering platforms for conlangers to publish high-quality work of interest to the community, raising awareness about conlanging amongst the general public, organizing work for professional conlangers and people in the entertainment industry interested in adding more depth to their alternative worlds, and providing a central place for reliable contacts and information to those seeking to learn more.

Most of what we do is non-profit work to serve our own community, though in this case we’re helping HBO by using our expertise to linguistically flesh out GRRM’s conworld.

EBW: Have any members of the LCS been hired as consultants before?

SE: About four or five of our members and associates. For example, Bill Welden worked on the Lord of the Rings movies (Bill was also one of the other finalists for this job, with a very impressive entry). Many (like David) have degrees in linguistics. Some are currently working on projects that are covered by NDAs. Most are like David, with a great deal of experience under their belt.

Suffice it to say that the people we work with for projects like this are the best of the best conlangers in the world. :-)

EBW: I haven’t read the books in a while, but I only remember a few Dothraki words—khalasar, khaleesi. About how big was the lexicon you started with?

David J. Peterson: There isn’t much extant material in the novels themselves—about thirty words, most of them names. There’s enough there, though, to give one a sense of what the language might sound like if it were fleshed out. That’s not to say there’s only one possible way the Dothraki language could have turned out (the initial proposals were quite divergent), but for me, there was enough material to figure out what direction I wanted to take.

EBW: Khal Drogo speaks to Dany in Westerosi some of the time. Did his mistakes in that language give you hints about the structure of Dothraki?

DJP: There are a number of elements that were borrowed in; for example, he refers to the Iron Throne as the “iron chair.” Consequently, it would be a bit strange for there to be a word like “throne” in Dothraki (if there were, presumably he would use the Westerosi/English word “throne”). One thing I did was further refine just how Dothraki-tinged English works. When I was at UCSD, I spent three years working under Grant Goodall for a class on first and second language acquisition, and I learned a lot about the types of errors that language learners make—what’s a likely error, and what isn’t—and I think based on the structure of Dothraki, I’ve come up with a pretty realistic not-quite-fluent “mode,” if you will, for native Dothraki speakers speaking English (and also vice-versa).

EBW: According to the press release, you drew inspiration from “Russian, Turkish, Estonian, Inuktitut, and Swahili.” Can you talk a bit about what you took from these languages?

DJP: Well, actually, the biggest inspiration for Dothraki was my own languages. Those languages were influenced by natural languages, though, so I listed those in the press release, since people will have heard of them. But, for example, the word building, and the way I expand the language, was inspired by Zhyler, a language of mine which was inspired by Turkish and Swahili. Come to think of it, an element of the noun case morphology was inspired directly by Russian—and also by my graduate morphology professor, Farrell Ackerman. Actually, his morphology class (which I took twice) influenced the way I look at language quite a bit (and is undoubtedly where my fascination with Estonian came from).

EBW: I vaguely recall that Turkish and Swahili are agglutinating (that is, they create new words by adding affixes). Does that mean Dothraki is agglutinating, too?

DJP: You’re right in your knowledge of Turkish and Swahili, but, no, I wouldn’t call Dothraki agglutinating. On the synthesis scale, Dothraki is closer to an inflectional language like Spanish or Latin. This doesn’t mean that affixes can’t pile on in certain circumstances (even English has this, e.g. “teach” → “teacher” → “teacherless” → “teacherlessly” [a word describing the teaching in an online class?]). There’s a chain of influence here. Swahili and Turkish heavily influenced Zhyler, and Zhyler influenced Dothraki. And while Zhyler is an agglutinating language, that aspect of the language didn’t influence Dothraki.

EBW: What made these languages in particular seem appropriate for Dothraki?

DJP: I didn’t try to select elements from certain languages that I felt were “appropriate,” or anything like that. I started with the material from A Song of Ice and Fire, and then tried to build out the language around that. I tried to imagine the circumstances of the Dothraki people, and that, combined with what I know about pre-industrialized languages and cultures, helped to determine what the language should look like, what vocabulary would be appropriate (what would be native, what would be derived, etc.).

EBW: What are some distinctive features of Dothraki, besides its subject-verb-object order? What might it sound like to an uneducated ear?

DJP: Hmm… You know, most people probably don’t really know what Arabic actually sounds like, so to an untrained ear, it might sound like Arabic. To someone who knows Arabic, it doesn’t. I tend to think of the sound as a mix between Arabic (minus the distinctive pharyngeals) and Spanish, due to the dental consonants.

A pharyngeal consonant is articulated with the very back of the tongue approaching the pharynx—the back of the throat. While English (like Dothraki) lacks any of these sounds, many Americans have been hearing the sound regularly since 1987. In The Simpsons, when Homer sees food he’s really excited about, his eyes roll into the back of his head, his tongue hangs out, he starts to drool, and he produces what can best be described as a kind of continuous pharyngeal fricative. If you try to mimic that sound, it almost feels like you’re choking. That “almost choking” feeling is right where your tongue needs to be to produce a pharyngeal consonant.

As for a dental consonant, we have two of them in English: the different “th” sounds in “think” and “this.” Those two sounds are fricatives, meaning that air passes continuously through the mouth, and the sounds can be held as long as one has air (analogous to “s” and “z”). In Dothraki, the “n,” “t,” and “d” sounds are pronounced with the tongue tip between the teeth, as it is for the “th” sound. It shouldn’t be too difficult to master for a native English speaker. Just try pronouncing the “t” in “taco” with the tip of your tongue between your teeth (and if you hold your breath while pronouncing the “t,” you’ll be coming remarkably close to a native Spanish pronunciation of the word).

Possibly the most noticeable feature about Dothraki is its use of circumfixes (inspired by Georgian, which I see I didn’t list up there) both for derivation and inflection. In English, we have suffixes (the “-s” in “writes”) and prefixes (the “re-” in “rewrite”), but, to my knowledge, no circumfixes. A circumfix is a simultaneous prefix and suffix (imagine if “rewrites” meant something, but neither “rewrite” nor “writes” meant anything in English). Circumfixes are used routinely in Georgian, but they’re pretty rare crosslinguistically.

Without giving too much away, I can show you some of the internal composition of athastokhdeveshizaroon. [Ed: according to the press release, this means “from nonsense,” and is the longest Dothraki word.] First, we’ll strip off the suffix (the last bit added), to get athastokhdeveshizar. With that done, the outer most bit is actually a circumfix. If you remove it, you get astokhdeveshi. So, in this instance, the circumfix is ath- -zar, and you’d write it thus.

EBW: What was the audition process for this project like?

SE: We had around 35 applicants for the job—I’d guess around half of them with a similar level of expertise to David’s. We wanted to give HBO the best possible proposals while giving everyone a fair chance.

Those who convinced me that they were capable of doing a conlanging gig (see current requirements) got an NDA contract and info about the actual job, and submitted several translations, a grammar writeup, and recordings for review. These submissions were reviewed by a panel of experts, fully double-blinded. Each application was evaluated for fidelity to canon, aesthetics, linguistic cohesiveness, originality, thoroughness/clarity, and of course a “fudge factor.”

The group settled on four finalists, with one back-up in case one of the four dropped out. Only after these choices were made were any of the reviewers told who the apps belonged to, and many were surprised when they found out. It was a lot of work for the reviewers, but it made certain that everything we showed to HBO was only the best of the best, and that it was well edited, organized, fleshed out, etc.

The four finalists were given the script for the pilot and some additional notes, and filled out the rest of the proposal: translations of all pilot material, etc. The finalists’ packages, together with some commentary and notes about the differences and highlights of each, were sent to the production team, who decided which one of the finalists got the job.

The HBO production team was wowed with what we sent; I think they hadn’t quite realized who they were dealing with. :-P

David is an LCS officer, but we treated him exactly the same as any other applicant, with full double blinding for review, so I’m confident that he’s a finalist only because he really is that good. All of them deserve kudos, though; it was some pretty tough competition, given some of the people we had apply.

That said, I think the process we used this time was unusually heavy. HBO was a very major client, and this project was worth some extra work to us just for the good publicity. In the future, most applications will be a significantly more streamlined process, to require less work from reviewers and applicants alike (while still, of course, giving the same quality final product).

DJP: From the point of view of an applicant, the process was intense. After all, I was competing with some of the biggest names in the conlanging world—people like John Quijada, Jan van Steenbergen, Rik Roots… Pretty much every name was a name I knew and respected (though there were a few new faces—people I came to know and respect). Some of the work produced during that application process is some of the most imaginative I’ve seen yet. I hope one day the creativity that was put into those proposals will make its way into other projects (under a different name, of course).

EBW: David, what first interested you in conlangs?

DJP: Like many conlangers, I didn’t know anything about conlangs until I was conlanging. Growing up, I never created languages (I didn’t even know that Klingon was a language, or that Tolkien had done anything but write books), but towards the end of high school, I became fascinated by foreign languages. In addition to AP Spanish, I took elementary German and started teaching myself Latin and French in my senior year. Then during my first year at Berkeley, I took Arabic, Russian, and Esperanto. It was my Esperanto class that introduced me to the idea that a language could be created. Then it was some time during my first introductory linguistics class that I “came up with” the idea that I could create my own language—not for international communication (as I supposed, at that point, all conlangs had been created for), but for personal use.

It was shortly after I started work on my first language that I discovered the online conlanging community via the Conlang listserv (whence the term “conlang”), and discovered, to my delight and horror, that I wasn’t alone (I had naively thought I’d hit on something no one had done before!). That was when my conlanging education began.

EBW: Can you tell me a bit about your own personal projects?

DJP: After I realized my first language was terrible (pretty much inevitable for a conlanger who starts without any outside knowledge of the craft), I gave it up, and started several new projects which survive to this day. My two largest are Zhyler and Kamakawi, the latter of which is, by a wide margin, my best. If it looks familiar, the language’s name came from Israel Kamakawiwo’ole (one of my favorite singers), and was inspired by Hawaiian. I think it really took off when I developed the orthography for it: a combination of a syllabary and a logographic system with over 600 glyphs. I’ve had a lot of fun working with it ever since.

I’ve developed a number of other languages, many of which are on my website, albeit with incomplete descriptions.

EBW: Can you explain what syllabaries and logographic systems are?

DJP: They’re different kinds of writing systems. English utilizes an alphabetic writing system, where we have glyphs (or letters) for each consonant and vowel in our phonology (more or less). A language like Japanese uses a syllabary, in which there are glyphs for each syllable. So, for example, we would write kimono with six letters in English: one for each consonant and vowel. In Japanese, if you were writing hatake (“field of crops”) in hiragana, you’d use three letters: はたけ, where は stands for ha, た stands for ta, and け stands for ke.

In a logographic system, or logography, a symbol stands for an entire word. While there are no pure logographic systems, many languages employ (or have previously employed) a limited number of logograms. Modern Japanese is one of these languages. In fact, the word hatake can be spelled as it is above, or can be spelled with a logogram (in Japanese, kanji). To write hatake, then, one could write はたけ or just 畑.

EBW: It sounds like GRRM is pretty pleased with your creation. Have you had any interaction with him directly?

SE: Yes, we have, and we’re looking at collaborating more in the future. Of course there are other issues involved, so we can’t comment on details or future possibilities other than to say that we’d be thrilled to work with HBO and/or GRRM on any of their future projects, and if fans want to see it happen they should make that very clear to HBO. ;-)

DJP: In fact, we’re exchanging e-mails right now. I’m totally jazzed (though I try to play it cool in the e-mails themselves)!

EBW: Were you a fan of the books before you got involved with the show?

DJP: It’s funny you should ask: Currently, A Clash of Kings is in my bathroom (resting right on top of Orlando Furioso). I had heard of the series before the show (my wife was a huge fan), and now all our copies are marked up with highlighter, a different color for each language.

EBW: Have you been or will you be on set at all as a dialogue coach?

DJP: I haven’t yet, but that may be because the status of a pilot is quite a bit different from the status of a greenlit show. I’m available, though, and so when it comes time to start filming again, we’ll see what happens. Thus far, I’ve provided them with recordings of every bit of dialogue to be used, so the actors have more to go by than just a transcription.

EBW: I’m guessing the writers don’t yet know all of the dialogue that will need to be written for the show. How will you start when they ask you for a word or phrase you haven’t invented yet?

DJP: First step: See if I’ve coined it already. The vocabulary keeps expanding (it’s probably up around 2,000 words), so I wouldn’t be surprised if I’ve covered a lot of bases. If I don’t have a word for it, though, there’s a process I go through. It’s simply not good enough to say, “I don’t have a word for brick, so the new word for brick will be blah.” It’s important to try to imagine the circumstances under which such a word would come to exist, and to decide if a new word will be derived from another word (perhaps via some standard derivational method), if it will be covered under a metaphorical extension of another word, if it will make for a sensible compound, or if it will warrant a brand new form. Which avenue I choose for a new lexical item depends largely on the word itself, and how I imagine it fitting into Dothraki society.

And, of course, the nice thing is that if you coin one word, it never stops there. Coining one word will often lead to a series of new words to kind of flesh out that corner of the lexicon. In this way, the language grows in an artificially organic way (if that oxymoron makes sense).

EBW: Any chance of a Dothraki-English dictionary being published?

SE: Yes, there’s definitely a chance. We’d be happy to do this. It’s not just our decision though; we need HBO’s signoff on any project like that. So again, if fans want to see this happen, they should tell HBO. :-)

EBW: Can you give us a few more words in Dothraki? How about something to make Dany proud—say, “the blood of the dragon,” or “I promise you a crown”?

DJP: Heh, heh… You know, I don’t even need to go to my dictionary to translate those.

“The blood of the dragon” is qoy zhavvorsi. [Ed: We know from the press release that “blood” is qoy, which means zhavvorsi must mean “of the dragon.”] Otherwise, for the time being, all I can do is direct you to the LCS Dothraki page. There’s some material there (including .mp3 files). We hope to announce plans for revealing more material soon, but we can’t do so yet.

Ellen B. Wright lives in New York, where she works in publishing and takes an excessive number of pictures. She’s already trying (and failing) to put together sentences in Dothraki, most of them involving the word zhavvorsi.


Back to the top of the page


This post is closed for comments.

Our Privacy Notice has been updated to explain how we use cookies, which you accept by continuing to use this website. To withdraw your consent, see Your Choices.