I am working on developing a fictitious language, and was wondering if there exists a commonly accepted set of phonemes which could be used.

I specifically don't want it to be similar to Indo-European languages, but do want it to be manageable for most readers ("speakers").

  • 4
    $\begingroup$ There are some general trends across languages regarding this (e.g. if you have click sounds you'll have at least ten of them), but no hard-and-fast rules. Have you read any conlanging guides or books? Zompist's "Language Construction Kit" and Peterson's "Art of Language Invention" are both excellent resources. Or if you join a conlanging group, people will generally be happy to help. $\endgroup$
    – Draconis
    Nov 17, 2017 at 16:56
  • 2
    $\begingroup$ Hi Ambluj, and welcome to Worldbuilding. I see you have accepted an answer already. Accepting an answer means you feel your question has been satisfactorily answered, and can reduce the likelihood of getting additional, possibly even better, answers. We typically recommend to wait at least 24 hours before accepting an answer to give people from different parts of the world a chance to see the question and compose answers. See Please wait at least 24 hours before Accepting an answer for a more thorough discussion on this. $\endgroup$
    – user
    Nov 17, 2017 at 18:17
  • $\begingroup$ Also, I would like to point out my old question For designing a vocal language for an intelligent species, how to decide on phonemes? which may be of interest to you as well. $\endgroup$
    – user
    Nov 17, 2017 at 18:18

7 Answers 7


Indo-European languages... have a common origin and common grammatical concepts; but they don't really have much of a common inventory of phonemes. A linguist, or an amateur who knows their basic linguistics, would understand the phrase "not similar to Indo-European languages" as referring to grammar, not phonology.

So what are the grammatical commonalities of Indo-European languages? (Note that English is typologically quite unlike other Indo-European languages. The history of the English language made it into a weakly inflected analytical, almost isolating, language.) Morphologically, they are fusional synthetic languages; most of them have grammatical gender; most have two or three numbers; most have at least a present tense and a past tense. Syntactically, they are nominative-accusative languages; most have articles; most exhibit grammatical agreement between nouns and adjectives and between subject and verb. Modern I-E languages vary between requiring strict word order (for example, English) and allowing almost completely free word order.

Follow the links to the categories listed in the previous paragraph to find non-Indo-European alternative options; for example, an agglutinative (like Turkic languages) ergative-absolutive (like Basque or Georgian) language with no grammatical gender, number, or tense would be very much unlike any I-E language.

You are absolutely free to invent any phonological system you want. But...

  • You cannot go wrong with the cardinal vowels /a/ /e/ /i/ /o/ /u/, the semivowels /j/ and /w/, one series of occlusives or stops (pick one series of /p/ /t/ /k/ and /b/ /d/ /g/), one rhotic consonant or one lateral (there are languages with lateral(s) and no rhotic(s) or vice-versa), the nazals /m/ and /n/ and the fricatives /s/ and /f/. Most people would be able to pronounce this set of phonemens or at least produce credible approximations.

  • For added variety you may introduce rounded vowels such as /ø/ and /y/ (most people can easily make them, even if they are not present in their language), diphthongs, a second series of occlusives (but beware that there is little universal agreement between languages with respect what exactly is the contrasting factor between the series of occlusives) and use both a rhotic and a lateral. Some languages allow both "simple" and "double" consonants (long or emphatic, depending on consonant type). Does yours?

  • Finally, consider the phonotactics of your language. Are consonant clusters allowed? Or maybe only some consonant clusters? How long can they be? Are they allowed in all positions? Is vocalic hiatus allowed? Can a word end in a consonant? Or in a cluster? Or in a vowel?

Have a great time developing your conlang!

  • $\begingroup$ I was referring to grammar when I used Indo-European, and thanks for the great response! $\endgroup$
    – Ambluj
    Nov 17, 2017 at 17:53

The UCLA Phonetics Lab has a public database of how sounds are used in different languages. There is also a nice website UI for the database, hosted by the University of Frankfurt. This contains a browser that can look up which sounds are used in which classes of languages.

For example, you can see that languages classified as Indo-European are:


You can then go see which sounds these languages use. For example, German uses these sounds:

p b k g pf f v s z S Z x m N l R h i: y: e: o/: E: a: u: o: I Y E "@ 4 U O j E) ai Oi au d ts n t

All of which have a special phonetic name associated with them. You could use this information to differentiate your new language in many ways. Here are a couple I thought of:

1. Use completely different sounds

This is the obvious way, but may be rather difficult as the Indo-European language class covers a very wide variety of sounds. You could look through and see which sounds are not used and make them the cornerstone of your new language. A potential downside to this is that written language often can sound many different ways. Even if you use different sounds, they may look similar on paper.

2. Use a smaller set of sounds

You might notice that Indo-European languages contain a very broad set of sounds. One way you could differentiate your language is to limit the amount of sounds. For example, Hawaiian uses a much smaller set:

p k ? m "n h "l w i E a "o u

Hawaiian's repetitive vowel sounds and consonants make it feel very different from German, even though they share much of the same phonemes. You could make your language feel unique by choosing a novel subset of sounds. This has the added benefit of allowing your language to be accessible to a Indo-European reader.

  • 2
    $\begingroup$ +1 for "Use a smaller set of sounds." A reduced set of consonants like in Hawaiian (or Rotokas, where some dialects only distinguish six consonant sounds), or a reduced set of vowel qualities like in Modern Standard Arabic, can make a language feel quite un-European without being unpronounceable. $\endgroup$
    – DLosc
    Nov 18, 2017 at 8:38

There aren't any commonly accepted sets of phonemes for constructed languages.

The cool thing about creating your own language is that the only rules you need to follow are your own. Often times people creating conlangs will take hints from existing language if they want to make a language that fits a particular purpose or mood. If you're wanting a language that is pronounceable for a particular audience then using phonemes they are already familiar with is probably best.

  • $\begingroup$ This may be somewhat of a different question, but to what extent can the familiarity of the alphabet used affect this accessibility to the language. I was leaning towards a syllabic writing system based of a mixture of symbols from the Arctic Circle and Babylonian writing. If I were to instead use a more Roman alphabet, is that then a tradeoff where the phonemes could become more foreign/complex? $\endgroup$
    – Ambluj
    Nov 17, 2017 at 15:46
  • 3
    $\begingroup$ If you're designing a language for accessibility, trying to create a more successful Esperanto, then this isn't the right forum for advice. If you're wanting to create a language for your work of fiction I wouldn't worry about accessibility. I'd worry about creating a compelling story. A good story with a bad conlang is still a good story. A bad story with a good conlang is something nobody will read. $\endgroup$
    – sphennings
    Nov 17, 2017 at 15:52
  • $\begingroup$ Funnily, there is a Standard Fantasy Phonology. Why does valar morghulis sound so Tolkinese? See my answer below: worldbuilding.stackexchange.com/questions/98131/… $\endgroup$ Mar 13, 2018 at 13:31

Lots of great answers here! Let me throw in a few thoughts...

Don't stress the alphabet. It can be jarring for if you start inserting deep-unicode symbols. Most readers can handle umlauts, accents, and possibly cedillas. ;D

Look at patterns of sounds. Here's a decent description of English phonology (sound inventory): https://en.wikipedia.org/wiki/English_phonology Not sure I like their use of more abstract fortis/lenis instead of voiced/unvoiced for English, but okay. To make your language exotic, play with the inventory; add some phonemes where English has gaps, and take some away. And add rules ... maybe in your language voiceless consonants become voiced between vowels (ie "happy" is not a valid word, but "habby" is). This is one of the reasons your speakers have funny accents. ;D

Look for consonant clusters which are "illegal" in English. Words don't start with (pronounced) "PT" or "KS", but Greek allows it.

Above all, be consistent. Languages have rules for syllable and word construction. If you make (or find) a set of rules, your language will sound a lot more realistic. Go here: https://www.vulgarlang.com/ to see an instant conlang generator!

Then think about how you are going to use the language. Is it a key plot point? A way to generate cool place/person names? Do place names mean something? You could have a local introduce characters to the magnificent cerulean city of Agorá-Prováton. "Wow," gasp the foreign travellers, "What does its name mean?" Then the guide could look down, and mumble "sheep market".

  • 1
    $\begingroup$ Apostrophes are typically okay, too. They are very often massively overused, though, so it might be a good idea to avoid them for that reason alone. $\endgroup$
    – user
    Nov 17, 2017 at 19:20

There is the IPA if you are looking into a way to denote the sounds human use in their languages.

Below, I interpret your question as "How to generate a few words in a distinctly alien/fictitious language?" There are two major possibilities.

Use an existing language

I have read a few "backstage" interviews of and secondary literature about successful established writers who had a similar problem in their novel. The solution was: take an existing rare language.

Examples I know include:

  • Take a Hungarian phone book, use it not only for person names, but for places and ideoms, like swearing.
  • Take a Mongolian dictionary and more or less translate your fantasy terms into it. Use them verbatim.

A more typical examples include magic spells in

  • Latin. Oh, hello there, HP fans.
  • Japanese. Just because they were verbatim in the original work and people don't bother to translate.
  • German. Just because it has the image of sounding brutal.

Design something new

It takes a lot of knowledge and work to design a proper conlang for your fiction. Tolkien is a prime example, another one would be Hiroyuki Morioka. The latter created a language that would be Japanese "purified from outside influences" and then developed independently for thousands of years. According to the lore, the actual speakers of this conlang had escaped the nazis who developed the language (hence the purity) and had to develop everything including written language themselves. So, it is sort of the French to Latin where Latin is the "purified" Japanese. But I digress.

Such complete language design is hard and requires some special knowledge. I have a far simpler idea that might work. In fact, I was so outraged that an absolute alien was called Samara in ME – Samara is a city name in Russia – that I came up with this idea some time ago.

The idea for a generator

You would need to program a small script or at least throw a dice quite a few times. Discern all the syllables you would like to have in a language, i.e. all that sound nice. Tear up some words from a dictionary in part or use a Katakana "alphabet". Pepper in some fricative/flap/nasal/affricate consonants or whatever should be characteristic sound of your local orcs. Then randomly put the syllables together to obtain words. Eventually, cut the "interim" vowels and alternate the spelling (bo-wu-ru -> bowel).

You would get words that are alien enough. Assign them a meaning freely.

Please still check if your freshly generated words not happen to mean something obscene or anything at all in any language. Google would help. (Occasionally even larger corporations fail to do so, but that's another story.)

  • 1
    $\begingroup$ Re generator, check out this site: vulgarlang.com It will auto-generate a conlang for you with an explanation of what it's doing. There's a free version, and a $10 version which gives you more options. $\endgroup$
    – akaioi
    Nov 17, 2017 at 23:44
  • $\begingroup$ Yeah, thanks. I saw it in another answer after I finished mine. It's pretty cool and I'd consider it a good writer's tool. $\endgroup$ Nov 18, 2017 at 11:15
  • 1
    $\begingroup$ "Specialist knowledge" - Tolkien's day job was Professor of Anglo-Saxon at Oxford. $\endgroup$ Nov 18, 2017 at 14:33
  • $\begingroup$ That's exactly what I was hinting on. But as Tolkien is a well-known example, I decided to bring on Morioka. I was quite sure he had some linguistics background too, but I just searched for it and found no proof. $\endgroup$ Nov 18, 2017 at 18:46

To answer the questions asked if there exists a commonly accepted set of phonemes which could be used: There is indeed kind of Standard fantasy phonology (SFP) that is very popular among fantasy writers. It can be roughly described as consisting of Spanish vowels and English consonants with the ach-Laut (IPA /x/ as in Scottish Loch, German Bach, Spanish junta, Esperanto ĥ) added. SFP also has typical phonotactic features such as frequent use of the vowel sequences ae and ie, only simplified allowed consonant clusters at all positions (initially, medially, and finally).


The Al Bhed from Final Fantasy X were given an amazingly unique sounding language with a simple alphabetic replacement code. Vowels were replaced by other vowels, and consonants for consonants, so the words remained pronounceable.

The word "important" translates as esbundyhd, pronounced: "eay-see-baeoon-deae-ha-de"

Here's an online translator: http://stefangagne.com/albhed.html

And amazingly, they did the same thing in Japanese and it worked just as well.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .