As is the case for many of us, I’m sure, my first fantasy love was the Lord of the Rings trilogy by J.R.R. Tolkien. Not only did I read the story itself over and over, I was also fascinated with the appendices attached. I could not help but marvel at the work Tolkien put into the background of his world and his obsession with detail, and nowhere was this more apparent than in his love of language and the tongues he created for the inhabitants of Middle Earth.

From this starting point I have always been interested to see how other writers go about creating new languages, and over the years I have seen examples of how to do it right and of how to do it very wrong indeed. As a writer I want to make sure that I fall in the former category, despite my natural inclination to laziness! I’m always keen to find new resources to help with my world building and make sure I keep learning, and I deeply admire those with a talent in this area.

So, it was a real pleasure to meet David Peterson (the creator of, amongst others, the Dothraki language for the HBO adaptation of A Game of Thrones) at WorldCon. An incredibly nice and humble guy, you would never know that he is one of the best in the business by him telling you, so I wanted to feature him on my blog to give others a chance see just how talented and knowledgeable he is.

It’s not often that a conlanger is asked to contribute to a writing blog. For a variety of reasons, serious language creation has always remained (and continues to remain) on the periphery of the artistic realm. Even today a high school student who writes “poetry” is more likely to be considered an artist than someone who’s been creating languages for fifty years (and there are several excellent conlangers both living and dead who fit that description).

But while some of the old prejudices remain, language creation has never been hotter. Starting with the Lord of the Rings film trilogy using authentic (or as authentic as possible) Quenya and Sindarin, enormously popular franchises like Avatar and Game of Thrones have profited from extensive use of one or more invented language. Their viability in popular media (and the visibility of these flagship franchises) has led many to consider making use of invented languages in their own works.

In this post, I’d like to give writers some practical advice on incorporating invented languages into their writing. Creating a full language may be out of the question for most writers, but hopefully understanding some of the issues behind language creation will be useful to all writers.

General Advice

Before going into specifics, I’d like to emphasize that in the case of language systems, less is more. It takes a lot of work to create an authentic language, and a great deal of expertise that comes only with study and practice. It’s easy to create a single conlang sentence, which, if translated directly into English without any other details, looks authentic. For example:

Brat danar glin kalag.
“Stay away from the swamp.”

There are potentially thousands of ways that the created sentence can mean what it means in English. For example, perhaps kalag means “stay away from”, brat danar is a compound that means “swamp”, and glin is an emphatic marker used to form commands. Who knows? There are any number of plausible interpretations. As more sentences are added, though, the possibilities decline rapidly:

Vor danar!
“Go away!”

All of a sudden we see danar appearing again in a sentence where it’s likely translated as “away”, which means the other word, vor, probably has to mean “go”, and the thousands of conlinguistic possibilities drop to the hundreds.

Add another sentence:

Mas bratuk danar glin kalag.
“We stayed away from the swamp.”

And now it’s becoming more and more likely that the “conlang” is just, in fact, a fancy way of speaking English, with some minimal changes (e.g. omission of the word “the”).

If one is prepared to invest the time to create a fully fleshed-out language, then, by all means, go for it! For those who just need the illusion of a language—something that could, perhaps, be expanded upon later if there were interest—it’s best not to get painted into a corner.

Sound Systems: The Place to Start

If there’s one aspect of language creation one should focus on if one needs a stand-in language for a novel, it’s sound systems (or phonology). The thing that most immediately distinguishes a language is the way in which all its words hang together. The best way to ensure one does this with a created language is to create a phonology to ensure all words share similar sounds and similar shapes.

Using English as an example, the following are all plausible English words:

• blork
• clund
• gettle
• sprile

The thing that makes them plausible English words is they all use English sounds and the sounds they use are arranged in patterns common to English. The following words, though, do not:

• ngsee
• dleh
• uzs
• jlemga

Each of the words above is composed of sounds that occur in English (e.g. the first word is just the “ng” from the word “song” placed right in front of the word “see”), but they’re arranged in ways that are completely impossible (or at least improbable) for English. If one were trying to create a fake English-like language, words from the first list would be far preferable to those in the second.

In creating a new language palette for a work of fiction, then, one ought to come up with a set of sounds (and they can differ from English. For example, let’s say there’s a language with no voiced stops [no “b”, “d” or “g”]) and then stick to them any time a word needs to be created. In addition, come up with a set of patterns in which the letters can occur. Here’s a sample language that has no voiced stops, no “r”, consonant clusters that allow only liquids after stops and fricatives, and syllables that end only in “n” or “l”:

• tlamen
• ilkun
• pyesu
• tikwil

Further levels of detail can be added (e.g. average word length, differences between word-internal and word-final codas, etc.), but the result is to produce word shapes that look like they come from the same language—and obviously so, since there likely won’t be a whole bunch of material for a reader to see.

Inflection

There are only a handful of ways that languages show inflectional morphology (e.g. pluralization, noun case, verb tense, etc.). The most common are:

• Affixation (including prefixes, suffixes, circumfixes and infixes).
• Word-internal change (e.g. “goose” > “geese” or “súbject” vs. “subjéct”).
• Prolix expression (e.g. “antiquated” > “more antiquated”)
• Suppletion (e.g. “is” vs. “were” vs. “am”).

Without creating actual grammar, one can give the illusion of grammar simply by employing one of these strategies on a key lexeme, e.g.:

Undiama tish hakhlor.                              Zvala dek sabindore lundiama.
“This peach is rotten.”                           “They devoured the peaches.”

The sentences above don’t reveal a lot about the grammar, but we see the presence of an l- prefix on what we would probably assume is the word for “peach”, and we see similarity in the suffixes for what may be the verbs of each sentence. Importantly, the translations still provide some latitude (for example, is the l- prefix a simple plural marker, or does it mark the accusative plural? Or is it perhaps simply a definite article, like “the”?), so we need say nothing definite, but the recurrence of recognizable morphology will give an otherwise fake conlang (i.e. a language that hasn’t actually been created, but is meant to appear as if it has been) a touch of realism.

The Myth of the Monolingual Planet

On Earth today we’ve got about 7,000 languages. While that number may be declining rapidly, it seems almost inconceivable (to me, at least) that there will come a point in time where every person on Earth speaks one language, with all the others having vanished. This, however, is precisely what we see in a wide variety of scifi and fantasy works: one people, one culture, one language.

Consider the case of English. English is spoken as a primary language in Canada, the US, the UK, Australia, New Zealand and parts of India. Imagine if one supposed that each of these countries was pretty much the same on account of the common language. And then move within one of those countries. Australians all speak Australian English (with no dialectal variation), so there, at least, is a region where pretty much everyone is the same, right?

Before one goes off to create 7,000 different languages for one’s fictional universe, though, it’s not that difficult to create the illusion of multilingualism. It’s rare to find a place where one and only one language is spoken. While there’s likely a dominant language, what are the minority languages spoken there? They can be referred to by name (even if they haven’t been developed), and can actually be reflected in proper and place names. Dropping these details in sparingly gives the illusion of depth, and results in a more authentic feel for one’s fictional world.

What’s In a Name?

Consider the following list of male names one might find in modern America:

• Thadeus
• Sam
• Jean-Paul
• Muhammad
• D’Brickashaw
• Daniel

What do they have in common? Almost nothing. As a result, one might, then, expect an equivalent list of male names in a fantasy novel to look like this:

• Sevander
• Vort
• Dhannïs-Shaer
• Zullebi
• Drîqz’z
• Forient

And they may look like that. But consider what the list of English names tells us. If we had to rank those names by most likely to be “American” (to the extent that’s a valid category), we’d probably start with “Sam” then go to “Daniel” then maybe “Thadeus” then probably “D’Brickashaw” (easily one of the coolest names ever invented). Why? Because the names themselves tell us something about their history. For example, we know that “Muhammad” (the world’s most common name) comes from the Arabic language, and is closely associated with Islam. We know that “Jean-Paul” is of French origin, and that a name like “D’Brickashaw” is uniquely American, but also not likely to have occurred before the year 1900.

Now let’s look back at the list of fantasy names. How realistic are they? That depends. Are these random names assigned to characters that have pretty much the same socio-cultural background? Their structure suggests not. If these are residents of the same region who share a common culture, it suggest that this is an area that’s seen a lot of immigration in its history. Does the detail given in the story back up or belie this interpretation?

As a final note, one often sees names such as the two listed below existing in the same system:

• Ambaliessa
• Dragonwind

That is, in this invented universe unconnected to our own, we have one name that comes from a non-English linguistic system, and another name that does. This doesn’t make a lot of sense. Certainly names are sometimes translated (consider “Sitting Bull”), but usually they all are, or they all aren’t. Consistency in this area is desirable. To those who pay attention to such things (more than one would guess), transgressions of this nature are roughly equivalent to the following:

Ambaliessa was a fair maid of the town of Parnilliat—a bit bookish, but uncommonly pretty. She would spend her days riding through the fields on her mare Semia, helping her father at the mill, and watching Gossip Girl.

You may protest, thinking Dragonwind isn’t that bad—and this may be true for some readers. For many, though, it’s a chink in the armor—a break from the alternate reality set up by the novel. For readers who like to get lost in the world of a novel, linguistic anomalies like this can be distracting.

The Phêÿqxh Language

There is a perception amongst linguistically aware readers that sci-fi and fantasy novels make liberal use of…”unique” romanization systems when rendering character and place names, often with the express intent of making the names look foreign. This practice often backfires when symbols are used in contradictory or counterintuitive ways. Consider the (admittedly cool-looking) name Xaro Xhoan Daxos. If pronounced in the most obvious way to an English speaker, the three x’s are pronounced three different ways in three different contexts (as [z] before a vowel in the first name; as [z] before a consonant in the second name; and as [ks] before a vowel in the third name).

While “it looks cool” is an adequate (albeit subjective) justification for most, one will run into problems if the fledgling system that gave birth to the name is ever expanded. One will often be forced to concede either that (a) certain characters are pronounced in contradictory fashions, (b) the romanization system is inconsistent, or (c) the character’s name isn’t actually pronounced in the way that’s most obvious to an English speaker.

Presuming that one’s readers are, for the most part, English speakers (if one is writing in English), I would propose that all names be spelled in such a way that their pronunciations should be obvious to the average speaker of English. As a result, certain practices should be avoided—namely:

• The use of any and all diacritics, save in situations where one really intends, for example, to include an umlauted vowel as used in German (e.g. “ö” or “ü”). Diacritics like the acute accent (“á”), the grave accent (“à”), and the circumflex (“â”) aren’t used in English, save in a few borrowings, and their functions are nebulous. Including them is bound to produce inconsistent results in the minds of different English-speaking readers.

• Inconsistent spelling systems. Unless the characters in the fictional universe of one’s story use the roman alphabet as their actual writing system (an unlikely coincidence if one’s world is unconnected to our own), there’s no reason to fabricate the inconsistencies found in natural language spelling systems that employ the roman alphabet (e.g. English, French, Vietnamese, etc.). To the extent possible, a policy of one letter = one sound should be employed.

• Non-English digraphs. A digraph is the use of two characters to convey a single sound. Thus, “th” in “thus” is a digraph, but “th” in “hothouse” is not. Certain combinations (e.g. “vh”, “xh”, “rh”, etc.) may evoke a certain aesthetic (and, indeed, may even prove useful in romanizing certain phonologies), but are likely to be misinterpreted or disregarded by many English speakers. (Incidentally, for the “g” sound in “genre”, I recommend “zh”, as it fits the following analogy rather nicely: s : z :: sh : zh.)

• Apostrophes. Apostrophes are used for contractions (e.g. cannot > can’t), glottal stops (like the ‘okina in Hawai‘ian), glottalized consonants (like the ejectives of Hausa) and occasionally pharyngeal consonants (like the first sound in the romanized spelling of the Arabic word for the Arabic language ‘arabiiya). Elsewise they are to be avoided at all costs.

Remember that in the fictional universe one sets up, it’s more than likely that the speakers of an imagined language will have their own unique writing system. The inconsistencies, irregularities and eccentricities one would find in a natural writing system should appear in that system, not the romanization used to convey the sounds of words and names in the language in one’s text.

Other Options

Of course, even this much may be more than a writer is willing to undertake, since the story comes first. Luckily there are literally thousands of conlangers all over the world who’ve been perfecting their craft for years—in some cases, decades. Many conlangers would love the opportunity to create a language for some sort of fictional work, whether it’s a big budget movie like Avatar, or a budding novelist’s very first book. If keeping a language or language sketch straight becomes too much of a hassle, I’d strongly encourage all writers to think about contracting out.

At present the best way to get a hold of a potential language creator is to go to the Language Creation Society’s Jobs Board. There one can post the details of a job and field responses from those who may be interested. Without exception, the best results come from those who have already put in their 10,000 hours.

***

It’s important to emphasize that the above is really just scratching the surface. Language is vast, and the possibilities are endless. If you’re interested in learning more about language creation, you can go to the Language Creation Society’s website, but one of the best ways to learn more about language, I’ve found, is simply to explore the grammar of a language you’ve never looked at. For this, Wikipedia is actually a great resource. Pick a language and type “[language name] grammar” into the search window, and see what comes up! Every language has something unique to teach us, and it’s never too late to start learning.

David Peterson is the creator of the Dothraki language for HBO’s Game of Thrones, and the alien language and culture consultant for SyFy’s upcoming drama series Defiance. He’s been creating language recreationally and professionally for twelve years, and currently serves as the president of the Language Creation Society.