In the language spoken by the Dothraki on the violent and popular HBO show Game of Thrones, the word for cat is keli. Reasonable, you might think. The show is based on the books written by George R.R. Martin, surely he can come up with whatever words he likes for the animals living in a universe he has dreamed up—that’s the fantasist’s prerogative. But Martin did not invent the word for the Dothraki cat. It’s named, in fact, after an individual real-world cat belonging to David J. Peterson, who has a web site which lists 13 separate languages of his own invention. Peterson has an M.A. in linguistics from the University of California, San Diego, and his bright blue, 1990s-chic web site has an alternate black-and-white version, in case you find it easier to read that way. Peterson is a conlanger, and he can do whatever he wants. 

“Conlang” is short for “constructed language,” which is just what it sounds like: a language that has been constructed. There are a lot of them, of various sorts. International auxiliary languages like Volapük, Esperanto, or Interlingua are one specific type of conlang. Invented to facilitate international communication during the great techno-utopian-modernist thought-boom of the last two centuries, they never got terribly popular. Conlangs do not necessarily have to be useful. As Peterson explains in his new book The Art of Language Invention, conlanging is an art as well as a science, something you might do for your own pleasure, as well as for the entertainment of others. He is a conlanger for hire—besides Game of Thrones, Peterson has also worked on Syfy’s Defiance, in which humans and aliens coexist in postapocalyptic Missouri—an artist who will put words into the mouths of the characters, words which are part of a fully functioning language.

You might ask, “How long has this been a job, and may I have it?” Fortunately, The Art of Language Invention answers this question, for it is both a textbook for the amateur conlanger—a guide to inventing your own language spoken by aliens or squid or submarine antelope or whatever—and a brief history of conlanging itself. And the story of conlanging is, as with so many other bodies of knowledge, the story of old-fashioned research inflated to surreal proportions by the internet’s bellows. Yes, you can be a professional conlanger, but the competition is stiff.


Once upon a time, in the first half of the last century, J.R.R. Tolkien invented not just a single language, but a whole family of languages which he fully fleshed out in private and then had some of his characters speak. For example, Sindarin and Quenya, two forms of Elvish spoken by different communities in his Middle-earth novels, are not only descended from a common ancestor, but are related to a whole slew of other languages. (The actual inscription on the ring from The Lord of the Rings is in Sauron’s own conlang, Black Speech, but using the Elvish lettering system Tengwar, in case you were wondering.) In 1931, Tolkien gave a lecture called “A Secret Vice” about language invention. I recommend elvish.org as a further resource for the curious reader. As the century wore on, a few people repeated the trick: Linguist Marc Okrand filled out the vocabulary of Klingon for Star Trek III: The Search for Spock, and British novelist Alan Garner gave the Skeksis, the villains of Jim Henson’s The Dark Crystal, their own language.

On July 29, 1991, the first message was sent to the Conlang Listserv. According to FrathWiki, a hub of information about conlangs and their surrounding cultures, “the list evolved from some informal email conversations among an early group of language enthusiasts.” Originally housed at Boston University, then at the Datalogisk Institut in Denmark, then at Brown University, the Listserv flourished. Of course, it was quickly splintered by tempers. Due to the “constant bickering” between advocates of different auxiliary languages (Esperanto and its sisters), certain members were banished to their own Auxlang Listserv in 1996. Nobody is allowed to advocate for a language on Conlang.

Next came non-English-speaking Conlang communities, such as Ideolengua, which held discussions in Spanish, as well as legions upon legions of interested young people, drawn to language creation less by Tolkien than the new stars of the 1990s. It’s hard to say exactly who these people were—Peterson doesn’t describe them as academic or nonacademic, but he himself doesn’t have a Ph.D, and this seems more like a hobby than a school thing. Some of Peterson’s favorite languages were created in this period, just for the fun of it: Teonaht, Skerre, Kēlen, Okuna (formerly Tokana), Brithenig, Ithkuil, Ayeri, and ámman îar. The emphasis on communication and entertainment, Peterson explains, was a huge factor in increasing the quality of the languages being created: Instead of being cloistered in a tower with no like-minded souls for company, these conlangers altered their languages just by talking to each other. Tolkien was good, but everyone can benefit from a little feedback.

Conlanging was a bit slow to gain broader acceptance, however, probably because it was an unbelievably nerdy pastime. It’s the sort of game little kids like to imagine they might play—I know I did—but only slightly saucer-eyed teenagers actually do. Auxiliary language fans, Peterson explains, “considered the practice counterproductive and silly,” while parents of the ’90s might even have thought there was something wrong with a child who started conlanging. 

Today, however, the internet heaves with baby conlangers and their work. Peterson rejoices in this fact, but has written this book for the very purpose of connecting those kids back to their dial-up forebears. The youth, he laments, know the word “conlang,” but have never heard of the Conlang Listserv. They know Na’vi from Avatar but it’s unlikely they have heard of Moten, a language created by Christophe Grandsire-Koevoets, the president of the Language Creation Society. This book is for them. It is a “sincere attempt to give new conlangers a place to start” by explaining what you need to know and what you need to do in order to construct your own language. Peterson wants the fledgling conlanger to ask, “What do I want to say with this new language that I can’t say in my native language—or any other language that currently exists?”


We are living in the golden age of conlang in popular culture. There are the many languages of Game of Thrones, Stargate SG-1’s Goa’uld, John Carter’s Barsoomian, and more. This is not to say auxiliary languages weren’t occasionally used for film work in times past. As Peterson notes, the 1966 William Shatner vehicle Incubus used Esperanto. His linguist friends, however, tell him the Esperanto spoken in it is “atrocious.” Apart from that, however, the big debut for conlangs on screen was 1974’s Land of the Lost. When the TV producers hired Victoria Fromkin of UCLA to create Paku, she became the first ever professional conlanger.

When Martin started writing A Song of Ice and Fire, he hadn’t actually developed Dothraki (the language with keli for cat) beyond a few words, let alone get anywhere near Tolkien’s achievement. So, Peterson was faced with a number of challenges. One related to matching up the sound of the language to the spelling. Peterson was determined to do justice both to fan expectations and Martin’s spellings (not least because fans had already tattooed them on their bodies) but mentions the “deviations” that keep him awake at night. A brilliantly, frustratingly basic example of this problem was Martin’s pronunciation of the word Dothraki, which was wrong. Peterson discovered that Martin pronounces the name of his invented kingdom and its language “DOTH-rock-ee,” putting the emphasis on the first syllable. “I’m glad I didn’t know about his pronunciation before I did my work,” Peterson admits, “I think I did right by the fans by going with what I think is the ‘usual’ ‘doth-ROCK-ee’pronunciation.” There’s also a big problem with the way we all say Khaleesi, but you will need to read Peterson’s book to understand exactly what it is.

The book gets truly strange, though, in its references to the sound qualities of languages that already exist. Dothraki is supposed to sound “harsh,” which means it uses the “ch” sound in “Loch Ness” a lot, so people think it sounds like German or Arabic. This impression of harshness is entirely subjective, and entirely culturally determined—the politics of which don’t get an enormous amount of examination from Peterson. Fair enough—that isn’t what his book is about. But Peterson does concede that sometimes his conlangs resemble real life languages, for specific reasons, as in the case of High Valyrian and its relation to Latin:

The High Valyrian language is meant to be the language of the old Valyrian Freehold and its vast empire on Essos in ancient days. The Valyrians conquered many lands—including the old Ghiscari empire—thousands of years before the action of the Song of Ice and Fire series, and their influence stretched all the way to the Isle of Dragonstone across the Narrow Sea. At some point in time a cataclysmic event destroyed the Valyrian Freehold, and the empire was wiped out. Many languages descended from the old High Valyrian language. These came to be known as the Bastard Valyrian tongues. The history of Valyria was modeled somewhat after the history of the Roman Empire, with its daughter languages descended from Latin. I wanted to honor this intention with High Valyrian without simply copying Latin, so I decided to take some cues from it without actually using it as a model.

Peterson is performing a very odd operation: He is taking a snapshot of European history and connecting it to the sound of Latin. A language that has mothered a whole bunch of contemporary tongues should be connected to an old empire: That empire enjoys a Golden Age, then a period of cultural benightedness, and out of this Dark Age rises languages labeled “bastard.” This is the snapshot we all grow up learning about the European past, a view of history that allows words like “barbarian” to describe people who lived a long time ago, but also to describe people who aren’t a part of Western civilization.

So, the language Peterson is using isn’t scientific, it’s an act of cultural interpretation. Which is, of course, fine, and a great reminder that linguistics is a subject that connects many different fields of knowledge. But when I see an advertisement for Game of Thrones, I see a cultural vision that is inaccurate, distorting, and bears a very specific relationship to the European past. It isn’t an innocent one. I see a Tolkien-inflected vision of the Middle Ages, with the sexual violence and warfare and ethnic stereotyping amped up to a pitch so cartoonishly horrendous that I can’t look at it. Of course, this is my professional bias—I’m an academic who works on medieval literature. By exactly the same token, however, I admire this book, written by the very author of my sadness.


Reading about the history of these imaginary languages is like seeing my own life and work played out in a parallel universe. In the dead languages I study—historical forms of English such as Old English, the language spoken by the Anglo-Saxons until about the twelfth century, and Middle English, which came after it—I often encounter words that defy understanding. Some of them have multiple meanings, and some don’t appear in any other text, so I can’t work out what they mean from their contexts. The texts are often missing big chunks, or they contain inexplicable phrases. Most of the famous poems in Old English—“Beowulf” included—exist in single, hand-copied manuscripts that have survived hundreds and hundreds of years and are often damaged. (The “Beowulf” manuscript was singed in a 1731 fire.) 

The conlanger does not have this problem: When words are missing, he just fills them in. Tolkien, the daddy of all conlangers, was also an Old English philologist. He brought a conlanging attitude to his reading of Old English; where there were gaps—as there are in the Old English poetic telling of Exodus, for example—Tolkien filled them in. He considered his grasp of Old English sufficiently expert that he almost treated it as his own conlang, subsequently flavoring the entire fantasy genre with a posthumous medievalism that has not been so subtly modulated as his own, and also apparently will not die.

It feels as though Peterson and I are Tolkien’s twin children, separated at birth. Our bodies of knowledge intersect significantly—we both have had to study phonetics and learn endless verb endings as well as the Great Vowel Shift—but I am desperate to recover the lost meanings of the past, while he delights in creating new poetries for the imaginary future. I also feel kinship with Peterson for the simple reason that we have both invested so much time and love into something so very uncool. The book has a few notes of nerdy prickliness in it, in fact. Of the conlanging crowd, he says that “a lifetime of negative feedback has left its mark on the community, which has been tolerant of praise, but allergic to criticism, constructive or otherwise.” Medievalists can be much the same. Like conlangers, we work in a field that can feel a bit marginal, a bit maligned by the literary studies mainstream. To me, though, it’s the most perfect art. And that, at the end of Peterson’s book, is the note that sells conlanging to me as something transcendently lovely:

I’ll tell you from personal experience that conlanging feels like an art to the conlanger. When I was a kid, all I did was draw. In high school, I picked up fiction, and then all I did was write. Finally in college I began conlanging, and all I did was conlang. For each stage, the drive—the motivation—was exactly the same: creation and expression. Inspiration comes from wherever it comes from and hits you, and then you’re conlanging. … I didn’t know a thing about conlanging until after I was doing it, so I can say for myself, it felt no different from any other art I pursued.

And medieval literature, which is likely much more dull and religious and less violent than an episode of Game of Thrones, feels like inspiration and creation and expression to me, too. “Ultimately,” Peterson says, “a language is nothing more than a system to encode meaning.” We share that opinion, at least.