THE (UNSPOKEN) NAMES, or, BUILD A FANTASY NAMING LANGUAGE IN ONE AFTERNOON
A.K. Larkwood, 7 February 2020
Coming up with a whole constructed language sounds like a lot of work, and don’t get me wrong: it is. There are people who make this a real artform, nuanced and detailed and linguistically rigorous. I’m sorry to say that’s not why I’m here. I’m here for one reason only and that’s to give some elves cool names.
Fantasy names are a topic dear to my heart. Maybe it’s from hearing too many uninformed jibes like “oh, those books where everyone is called Blarg and Klarg”. Maybe it’s from my perverse love of every scene in Lord of the Rings where Gandalf turns up in some new place and gets a new name enunciated at him (“Mithrandir!!!”). Either way, the names you use in a work of fantasy undeniably lend a certain flavour to the world, and it can be fun to come up with a consistent way of generating them. So, I thought the launch of my novel The Unspoken Name would be a good time to explain how I do it.
FIRST THINGS FIRST
Let me say that you don’t have to follow this method. There are a lot of approaches – adapting modern names, using names from a historical period, developing a naming system which works in some completely different way (see for instance Arkady Martine’s A Memory Called Empire) – and any of them can work just fine.
This is a strategy I’ve used to develop distinctive naming languages with a certain baseline of phonetic coherency. That baseline is a matter of my personal taste and it is open to you to adjust up or down depending on whether you remotely care.
Some brief disclaimers:
this is going to be English-centric since it’s the only language I speak fluently – apologies.
the end goal here is something broadly readable to a non-linguist so I am using English spelling rather than the International Phonetic Alphabet.
in case it’s not obvious, I’m not a real linguist - at best an enthusiastic amateur!
PRINCIPLES
It can help to think about some guiding principles for the naming language you want to develop, depending on the story you’re telling. If you’re writing about a very stratified society you might want to reflect that in a naming system which has lots of status markers, for instance. Or you might be writing about an alien species and want to use features which aren't found in human language.
Here are the principles I had mind when I was developing the character names for The Unspoken Name:
1) CLARITY. I wanted the names to feel at least somewhat user-friendly to a reader accustomed to English.
2) NOVELTY. There are no “fantasy counterpart cultures” in the book. The three major setting - Oshaar, Qarsazh and Tlaanthothe - aren’t based on any real-world societies of the present or past, so I didn’t want to use or modify names from existing languages.
3) DISTINCTIVENESS. The distinctions between these three cultures are meaningful and relevant to the action. I wanted to make sure the reader could distinguish them fairly readily by the “feel” of the names. I needed to come up with three naming languages which felt immediately different - these ended up being Oshaarun, Qarsazhi and Tlaanthothei.
A note on distinctiveness
Obviously, this is a brutal simplification of the way things work in the real world, where naming practices are complex and granular and often freighted with meaning. You only have to look at baby name lists to see a sample of the range we’re working with in English.
But there’s often a trade-off in worldbuilding between subtlety and clarity. Modelling any one feature of an imaginary world with a ‘realistic’ degree of nuance and detail asks a huge amount of the reader’s ability to digest information - best saved for the aspects which really matter to the work.
Here, since I was asking the reader to get to grips with several invented countries for which they had no frame of reference, sharply marking that distinction with names was an easy way to make things clearer:
Oshaarun: Tsuru, Csajad, Najalwe, Ejarwa, Ankhaad, Psamag, Sharai
Qarsazhi: Riqpi Qilkhiri, Bazha Sadhmira, Atqa Raxali, Vanchi Tsavaydra
Tlaanthothei: Essanthe Lathraai, Kalene Martela, Keleiros Lenarai, Akaro Klareia
If I’ve done my job, it should be clear that each has a distinct flavour and style.
But how do you make that happen?
There are four steps.
you learn a little bit about phonetics
you don’t bother learning much more about phonetics
you do a small amount of thinking
and then you cheat.
LEARN A LITTLE BIT ABOUT PHONETICS
Consonants
Here’s an extremely brief rundown of how consonants work.
There are several different types of consonants. The ones I’ll be considering here are:
stops. where the flow of air is briefly stopped to make the sound. all the consonants in pat, tack, keg, bed are stops.
fricatives. where the flow of air is only partially blocked to make the sound. all the consonants in shave, size, this, thief are fricatives.
affricates. where a stop and a fricative get together. the ch in church is an affricate composed of t plus sh, for instance. Not all models treat these as individual consonants in their own right and it may be easier to think of them as consonant clusters (which we will be addressing shortly).
liquids and glides. where the flow of air is hardly stopped at all. all the consonants in rule are liquids and all the consonants in way are glides, but for these purposes I tend to treat them as one category.
nasals. ones you can hum. all the consonants in meaning are nasals (usually – there are British accents which hit the g at the end a bit harder).
An important property of consonants is voicing. This is the difference between the first sounds in bet and pet, or call and gall. If you say these pairs out loud you’ll be able to feel the difference in vibration. Voiced consonants sound in your vocal cords. If you whisper, every sound becomes unvoiced.
Finally, let’s look at place of articulation. If you say the stops p t k out loud, you’ll notice that the point of contact changes for each one, from your lips, to just above your teeth, to the very back of the roof of your mouth. The combination of type and place of articulation is what gives the sound its character (p and m have the same place of articulation but a different type, for instance). There are a lot of places of articulation and it’s easier to digest them in table format:
Place of articulation: | Bilabial (lips) | Labiodental (lips and teeth) | Interdental (teeth) | Alveolar (just above teeth) | Palatal (roof of mouth) | Velar (roof of mouth but further back) | Glottal (right at the back) |
Stop (unvoiced) | p | t | k | glottal stop | |||
Stop (voiced) | b | d | g | ||||
Fricative (unvoiced) | f | th (in thin) | s | sh | h | ||
Fricative (voiced) | v | th (in this) | z | zh (middle sound in treasure) | |||
Affricate (unvoiced) | ch | ||||||
Affricate (voiced) | j (in judge) | ||||||
Nasal | m | n | ng (in sing) | ||||
Liquid | l | r | |||||
Glide | w | y (in yam) |
There’s a lot more detail and complication you can learn about if you’re interested, but this is a decent place to start, a rough overview of the range of consonants in English. Other languages have different ranges - you can see there are some obvious big gaps in the table which can be filled in with other sounds.
STOP LEARNING ABOUT PHONETICS
This is now all you need to know about phonetics to make a fairly decent fantasy naming scheme.
You may be thinking, what about vowels? They have a lot of fun properties like length, roundedness, they can be palatalised or nasalised, or you can have pitch and tone, and what about stress -
But we’re working in a written medium, and the problem with vowels is that English spelling makes a mockery of them. Spoken English in all its variants has far more vowels than the alphabet can cope with and as far as I can tell we’ve given up trying. In my accent, the vowel sounds in “war” and “brought” are about the same, while “cat” and “rough” are doing completely different things, for instance.
Of course you could come up with an elegant, consistent system of vowels for your imaginary language, and more power to you if you do so. I tried to do this a few times. What I mostly learned was that I’ve been so damaged by reading English all my life, I’d just ignore my own rules and impose whatever pronunciation seemed good. It seemed a fair bet that readers would do the same, and it was a lot of effort to implement, so I stopped.
I’d say just pick whatever vowels you like and use them however you want.
Here was my thought process on the vowels in Oshaarun:
“I’ll just use a e i o u and see how it pans out. No, wait, I’ll skip the i. Damn, but I already have one name with ai that I really like. Maybe it's only allowed in that context. Let’s have a, e, o, u, ai and a long aa, that's fun.”
A SMALL AMOUNT OF THINKING
At this point it’s time to make a few decisions which will do a great deal to affect the sound and feel of the naming language.
Pick your consonants
You can do this at random, or you can add or delete rows or columns from the table. (You might want to print it out). Do as you will - most languages have the simple stops p, t and k, but not all. For what it’s worth, I’ve had some interesting results simply taking the English table and making some choice deletions.
In the likely event that you already have some character or place names in mind, you can do some reverse-engineering - write down as many names as you have and make yourself a list of consonants that occur. You may find that they already make an interesting pattern in the table.
To find out about sounds that aren't in the basic English table, I recommend visiting the Interactive IPA Chart, which has recordings of all the sounds in the International Phonetic Alphabet arranged in a similar way to the table - but bear in mind that you'll need to find some way to represent them on the page and convey that information to the reader. This is pretty easy for affricates like ts, and in lots of cases there's a familiar way to do it, like the Spanish ñ, so don't be put off if there's a sound you want to use that’s uncommon in English.
You can also add some restrictions about where certain sounds occur. English has some affricates which often appear at the beginnings of words (“church”, “judge”) but others are rare outside loanwords (“tsetse”). I quite like the affricates ts, ks, ps, so I decided to use them all over the place in Oshaarun, notably in the main character’s first name (“Csorwe”).
For reference, here is the consonant table for Oshaarun:
Place of articulation: | Bilabial (lips) | Labiodental (lips and teeth) | Interdental (teeth) | Alveolar (just above teeth) | Palatal (roof of mouth) | Velar (roof of mouth but further back) | Glottal (right at the back) |
Stop (unvoiced) | p | t | c | ||||
Stop (voiced) | b | d | g | ||||
Fricative (unvoiced) | s | sh | kh | ||||
Fricative (voiced) | |||||||
Affricate (unvoiced) | ps | ts | ch | cs | |||
Affricate (voiced) | j | ||||||
Nasal | m | n | ng | ||||
Liquid | l | r | |||||
Glide | w |
As you can see, it's not very different at all from the English table - this is really just a starting point. The magic happens in the next two steps.
Syllable structure
Once you've chosen your range of sounds you need to think about how they'll be arranged.
As an example, let's suppose we’ve decided to use only the consonants p, t, k, m, n and r. The only vowels in this language are a and i.
You can write this down as:
C = p t k m n r
V = a i
Even with this restricted palette of sounds we can produce some varied results using different syllable structures.
One basic syllable structure is CVC - that is to say, every syllable must consist of a vowel surrounded by two consonants. pat. kim. ran. nir.
Maybe you decide the last consonant is optional - it can appear, but doesn't have to. ta. nik. ri. nam. You can note this down as CV(C).
Or maybe both consonants are optional. ip. a. min. ka. This would be (C)V(C).
You can get more complicated. Maybe you want the first consonant to be optional, but your syllables to end with a vowel or “r”: i. ta. mir. ar. You could note this as (C)V(r).
Maybe you only want your syllables to begin with an unvoiced stop. In this case you could have some other letter such as U stand for that series of sounds:
C = p t k m n r
V = a i
U = p t k
And then you can note the structure as UV(C) - pan. pi. tim. ka.
English has a pretty free and easy syllable structure. You can have a whole mess of consonants either side of the vowel, and several vowels in between them if you want. Other languages have more or less restrictive schemes so feel free to choose whatever rules you like.
You might want to think about how many syllables you'll have per word and how that will work for names. Some languages have lots of monosyllables. Some have lots of polysyllables. Mixing this up is an easy way to distinguish one from another on the page.
A useful tool for experimenting with syllable structure is Mark Rosenfelder’s Gen. It can be hard to see intuitively what effect different syllable structures will have, but the tool will let you generate some text using a given structure so you can play around and see what works for you.
Consonant clusters
All languages have rules about which consonants can appear next to each other, and when. In English some pretty gnarly consonant clusters are allowed - consider the first sounds in “stripe” “split” “scrape”.
But there is method here - the formula is s, followed by an unvoiced stop, followed by a liquid.
Some clusters appear only in certain places. English often has “nd” and “ng” at the end of a syllable but rarely at the beginning (except for loanwords). Other clusters tend to only appear at the beginning of a syllable.
Thinking about what clusters you'll use does a lot to shape the feeling of the language. It can be a bit time-consuming to go through every possible combination but it's worth at least getting a feel for which clusters are characteristic.
Let’s go back to our example language and add some more sounds.
C = p b t d k g m n r s z
V = a i
Some example rules you could use with this set:
[stop] + r: pra, atr, bri, igr
[stop] + [fricative] (but only at the beginning of the syllable): tsa, dzi, bza, ksi (NB you could also treat these as affricates if they're really common in your language - whatever approach suits you better)
[stop] + [stop] (but only between vowels): ipta, igdi, akba, adbi
[nasal] + [unvoiced stop]: npa, int, nki, amp
As a needlessly complicated example, I decided that, in Oshaarun, [unvoiced stop] + w would be allowed, but if there was a nasal before the stop it would become voiced (eg “Cweren”, “Angwennad”).
Play around and see what seems to work. The Gen tool can be useful for this. You don't have to do anything fancy - just allowing, say, [stop] + r or [nasal] + [stop] will give you a good variety. Again, if you have some existing names you like, have a look and see what clusters already exist - you may find that they suggest some rules you can stick with.
For bonus points you can think about ease of pronunciation. Sounds often get skipped or altered if they're hard to say - be honest, do you actually pronounce both the f and the th in the word “twelfth”? This can have an effect on the clusters you decide to use. For example, it’s hard to pronounce a nasal and a stop together if they have different places of articulation. umber and under are easy to pronounce. umder or unber would be trickier, though not impossible - consider “unbearable” for instance.
Wider questions to consider
Do you want your names to mark gender, whatever that means in your world?
Do you want your names to include status markers like titles or honorifics?
Do your people use family names? What about nicknames?
These are really all worldbuilding questions but they can be fun to think about.
Once you’re happy with your range of sounds, syllable structure and consonant clusters, you have done the hard part...
AND NOW YOU CHEAT
As well as not having enough letters to handle all the sounds, English spelling has a few sounds where there are multiple ways to represent the same sound. This is probably a nightmare if you're learning the language but for us it is a feature not a bug.
For instance, “c” “k” and “q” all represent distinct sounds in some languages, but in English they’re basically the same. If you’re working with multiple languages and you want to distinguish them on the page, you can do worse than pick one variant for each and stick with it.
(I did this exact thing: Oshaarun uses “c”, Tlaanthothei uses “k” and Qarsazhi uses “q” for what’s essentially the same sound)
You can also think about how you want to represent consonant clusters. “cs”, “ks”, “x”? “ch” or “tsh”? Vowel sounds give you even more latitude if you want to tangle with them – “whey”, “fray”, “Mae”, “lei”, etc.
Once you have all this - a list of sounds, some rules about how they’ll be assembled, and perhaps some rules about how they'll be represented on the page, you're basically ready to go. You can plug them into Gen or work some out for yourself. The result should hopefully be pleasantly consistent and you can come back to it whenever you need to name a new character. (It’s also a good starting point if you’ve at this point been tempted into the vortex and suddenly feel the need to start developing verb forms. This was of course my purpose all along).
FURTHER READING
If you’d like to see how this plays out in fiction, check out my book The Unspoken Name, out in February from Tor Books.
The Language Construction Kit is responsible for getting me into this in the first place and goes into everything I’ve described in far more detail - a good place to start if you want to go further than the bare-minimum approach I’ve used.
Language Log is an interesting linguistics blog.
Lingthusiasm is a fun podcast about linguistics.
Vulgar is an online tool for generating whole new languages to use.
The Rinkworks name generator is a simpler, but still very tweakable, option if this whole endeavour seems like too much hassle.
The interactive IPA chart is a useful tool if you want to start using the International Phonetic Alphabet and get a sense of the sounds which exist in pockets of consonant space which I haven’t covered.