Conlang Romanization

[shmittygoatman]

Hi guys! I'm fairly new here, but I think this is a great idea for a conlanging community. Since I'm fairly recent, I thought I would go ahead and ask a recent question that has been on my mind.

I often have trouble romanizing my conlangs (writing down the different sounds in the Roman alphabet). My languages often have sounds that would either be hard to write in the Roman alphabet, or nearly impossible. One of the main reasons I make conlangs is for my sister. She writes a lot of fantasy and sci-fi books, and often needs different cultures or species to have different languages. I have made languages for Orcs, Dwarves, Elves, etc. One of the problems I run into is that I can't write these languages down in a way that would make sense to the reader.

I've seen that some languages, like Klingon, use uppercase v. lowercase letters to indicate different sounds. The problem with that system is that my English-speaking brain naturally views that differently, so it looks to me like the speaker's voice is constantly moving up and down.

Anyways, any help would be appreciated. Thank you!

[eldin raigmore]

Z-SAMPA, and CXS or Conlang X-SAMPA, are two ways of writing using ASCII characters, which are mostly Roman alphabet characters (uppercase and lowercase having different meanings).

On this CBBoard we used to prefer Z-SAMPA until http://ipa.typeit.org/full/ came along.

If you don't want to use non-alphabetic characters, you're limited to about 52 phonemes.
If you also don't want uppercase and lowercase versions of the same letter to have different meanings, you're limited to about 26 phonemes.
Unless you either use diacritical marks, or use digraphs.

If you had four diacritical marks and any combination of the four (including "none of them" and "all four of them" and "any one of them") could be applied to any letter (whether uppercase or lowercase), you could have 16*26 = 416 different phoneme-symbols.

If you use digraphs, there would be 676 of them that you could use; but you'd need to come up with some way to make sure the reader could reliably distinguish a digraph from merely two letters used one after the other but independently.
You'd also need to make sure the reader could tell whether three letters in a row, say "xyz",
should be read as the digraph xy followed by the letter z "xy z",
or the letter x followed by the digraph yz "x yz",
or the three letters x and y and z "x y z".

[Xonen]

I often have trouble romanizing my conlangs (writing down the different sounds in the Roman alphabet). My languages often have sounds that would either be hard to write in the Roman alphabet, or nearly impossible.

Oh, it's possible to write down pretty much anything in the Roman alphabet, you just have to get a bit creative sometimes. See, say, Xhosa for starters.

[shmittygoatman]

I often have trouble romanizing my conlangs (writing down the different sounds in the Roman alphabet). My languages often have sounds that would either be hard to write in the Roman alphabet, or nearly impossible.

Oh, it's possible to write down pretty much anything in the Roman alphabet, you just have to get a bit creative sometimes. See, say, Xhosa for starters.

While that may be true, it would be hard to communicate that to the reader of the language. For example, if I had an aspirated voiceless dental plosive, I could write that using /t/. However, if I already have a /t/, I might have to use /t'/, which gives not much clue to how it is pronounced, or what makes it different from just regular /t/.

[Xing]

There are different parameters in romanisation, that could pull in different directions. Which direction one should follow, can depend on the goal and the intended audience of the romanisation. Is the romanisation meant to give non-natives learners and/or occasional readers a rough estimate of the pronunciation, or is it meant to be used 'in-language'? If the romanisation is mainly meant for learners of the language, it might a good idea to make it as unambiguous as possible. For instance, it might be good to have some way of telling whether a consonant is part of the coda of syllable one, or the onset of syllable 2, or whether it's part a digraph or not. (For a 'native' speakers - whether real or fictional - minor ambiguities would typically not be a problem.)

If the romanisation is primarily targeted towards occasional readers, one might choose to indicate major allophonic variations - even if this kind of marking might be superfluous from an 'in-language' perspective. Say that, in a language, [t] and are allophones - occur before high vowels, [t] elsewhere. You could then choose to indicate this difference in the romanisation: <si su>, but <te ta to>. This would not be necessary from a 'in-language' point of view - since the distribution is predictable. But it can be still be useful in that it gives a hint of the pronunciation for 'outsiders', who might not be familiar with the phonological rules in the language.

[Dormouse559]

If your conlang is intended for an uninitiated audience, perhaps avoid straying too far from sounds/phonemic distinctions that they know how to pronounce or approximate. Additionally, think about making the romanization similar to that of a language they're likely to have experience with. It may in the end make the phonologies a bit less imaginative, but it will make them more accessible, and suggesting pronunciation should become easier.

[eldin raigmore]

What everyone else (including shmittygoatman) has said up 'til now, plus:

... if I had an aspirated voiceless dental plosive, I could write that using /t/. However, if I already have a /t/, I might have to use /t'/, which gives not much clue to how it is pronounced, or what makes it different from just regular /t/.

How about /th/ ?

And there are other possibilities; to name just three, /th/, /ht/, /ht/.

[shmittygoatman]

Thanks all you guys for your help. I guess I was looking for too perfect a method for writing my languages down.

Oh, and also:

What everyone else (including shmittygoatman) has said up 'til now, plus:

... if I had an aspirated voiceless dental plosive, I could write that using /t/. However, if I already have a /t/, I might have to use /t'/, which gives not much clue to how it is pronounced, or what makes it different from just regular /t/.

How about /th/ ?

And there are other possibilities; to name just three, /th/, /ht/, /ht/.

I could write that, but I run into two problems. First, using superscripts and subscripts may be useful for someone who understands phonology, but people who don't understand it will often be thrown off or distracted by it. Also, (in English anyway) /th/ of the English digraph [th], which is pronounced completely differently than /th/. I still have to make a mental adjustment every time I come across it.

Anyways, thanks for all your help.

[Thakowsaizmu]

If the alien languages are for a book or some such, honestly, most people are going to skim right over it, save for the random relevant word here and there. For example, <Jeddak> means "Emperor". So looking to perfectly romanize an alien language for an English speaking audience isn't likely to happen. In a way, that's a good thing. Create your romanization without worrying too much about making sure that the reader perfectly pronounces the word, but for those with earnest interest, create an appendix in the back that'll help to explain further, like many novels do.

[Thrice Xandvii]

A big to what Thakowsaizmu said. She said pretty much exactly what I would have suggested... but better.

[Xonen]

What everyone else (including shmittygoatman) has said up 'til now, plus:

... if I had an aspirated voiceless dental plosive, I could write that using /t/. However, if I already have a /t/, I might have to use /t'/, which gives not much clue to how it is pronounced, or what makes it different from just regular /t/.

How about /th/ ?

And there are other possibilities; to name just three, /th/, /ht/, /ht/.

Are these meant to be phonemic transcriptions or romanizations? The use of slashes suggests the former, but there are already standard ways of doing that, and I don't get the sense that's what's being asked here. And for romanizations, I'd stay away from superscripts and subscripts; makes text kinda ugly and difficult to read if there are a lot of those. Although that may be just my personal opinion.

---

If the alien languages are for a book or some such, honestly, most people are going to skim right over it, save for the random relevant word here and there. For example, <Jeddak> means "Emperor". So looking to perfectly romanize an alien language for an English speaking audience isn't likely to happen. In a way, that's a good thing. Create your romanization without worrying too much about making sure that the reader perfectly pronounces the word, but for those with earnest interest, create an appendix in the back that'll help to explain further, like many novels do.

I'll third this. If the problem is that you're looking for ways of representing sounds that don't occur in English so that it would get monolingual speakers of English with no linguistic training to pronounce them correctly, then yes, that is indeed quite probably impossible. But it's not due to the limitations of the Latin alphabet, and there's pretty much nothing you can do about it. So no use getting hung up on it. Better to give the language geeks something genuinely interesting, than to dumb it down for those who'll mostly ignore the whole thing anyway.

[Nortaneous]

Here are some orthographies of some natural languages.

Southern Qiang:
/pʰ p b tʰ t d kʰ k g qʰ q/ <p b bb t d dd k g gg kv gv>
/tsʰ ts dz tʂʰ tʂ dʐ tɕʰ tɕ dʑ/ <c z zz ch zh dh q j jj>
/f s z ʂ ʐ ɕ (ʑ) x (ɣ) χ ʁ h ɦ/ <f s ss sh rr x xx v vv hv vh>
/m n ɲ ŋ/ <m n ny ng>
/ɬ l (w j)/ <lh l w y)
/a ɑ e ə o i y u/ <ae a ea e o i iu u>
V˞ <Vr>

Gvearlozhuvva, mutula muyugvu rrgguessi weai, ssebleea eqekvarrei. Me nya hhlu yezhnguenyi, ssuamefu orrgguda ibivuanyi, vhomuhjuweai. Sdeageda mi bearrres nguevuanyi, me nya hhllu lessi idhiji, nyujugu zzuavvl eahshea vheaibbinyi, vvugvdawu vhohhlu, zzuahhl yehshe eazheadubdu, nyissssi ogvbi bes, zzuahhl yehshe vhohhlusda, kvuad eazheadubueai, vhazeinyigea, me nya hhlu lessi ogvbi vhobei.
(the <rrr> and <ssss> are not typos: <bearrres> is /pe˞ʐəs/ and <nyissssi> is /ɲizzi/.)

Somali:
/b t d tʃ ɖ k g q ʔ/ <b t d j dh k g q '>
/f s ʃ x ħ ʕ h/ <f s sh kh x c h>
/m n l r j w/ <m n l r y w>
/a e o i u/ <a e o i u>

Barandhaha oo ah khudaar midho wareegsan oo midabo badan leh, inta badan huruud cadaan ku dhex jiro, ayaa ka mid ah khudaarta lagu daro cuntada ama kaligiis la karsado. Midhaha geedka barandhadu waxay ka baxdaa carada dhulka hoosteeda. Khudaarta maalin walba dadku isticmaalaan waxaa ka mid ah: basasha, tamaandhada, barandhaha, saladhka, toonta, kaarootka, besbaaska, iyo kuwo kale oo badan.

Seri:
/p t k kʷ ʔ/ <p t c~qu cö h>
/ɸ s ɬ ʃ x xʷ χ χʷ/ <f s l z j jö x xö>
/m n j (l r)/ <m n y ḻ r>

Ox tpacta ma, xiica coi yeen iicp quiij cmiique caii Oficj Cooil hapah quih yeen iicp tiij, toc cötiij ma, hacat timoca áno tafp, iixl cah itexl, itahit, toc cöyiihtim. Ox tpacta ma, ziix cmiique caii Oficj Cooil tintica itcaail, iiqui mpanzx xah mimoz xo quiha ha. Canoaa quih quiha hapah quih ihiha, taax oo cöquiha ha yax, iixöt quih caacoj hipcom cömiifp xo iixöt hipcom cöitacoxot, yoque. Haaco tama ma, aama. Pooixaj ta, ox impacta ha.

Nuosu:
/pʰ p b ⁿb tʰ t d ⁿd kʰ k g ⁿg / <p b bb mb t d dd nd k g gg mg>
/tsʰ ts dz ⁿdz tʂʰ tʂ dʐ ⁿdʐ tɕʰ tɕ dʑ ⁿdʑ/ <c z zz nz ch zh rr nr q j jj nj>
/f v s z ʂ ʐ ɕ ʑ x ɣ h/ <f v s ss sh r x y h w hx>
/m̥ m n̥ n ɲ ŋ/ <hm m hn n ny ng>
/a̠ ɯ ɛ̠ i ɔ̠ u ɿ̠ ɿ ʮ̠ ʮ/ <a e ie i uo o yr y ur u>
(underline = retraction, ɿ ʮ are the apical and labial fricated vowels. ɿ is z̩ and ʮ is v̩~ʙ̩.)
/˥ ˧˨ ˧ ˨˩/ <-t -x -0 -p>

Nbo ma mu viex jjux jjo, nzy ddu i qix jjy yyx mu jjo sat. Nbo wox ngop mge si nip bbop hxie nyi jjo, ddix ap bbop hmap zyt hnip mop mu jjo tat xi.

Kabyle:
/b t d tˤ dˤ ts dz tʃ dʒ k g q/ <b t d ṭ ḍ tt zz č ǧ k g q>
/f s z sˤ zˤ ʃ ʒ χ ʁ ħ ʕ h/ <f s z ṣ ẓ c j x ɣ ḥ ɛ h>
/m n l r rˤ j w/ <m n l r ṛ y w>
/a ə i u/ <a e i u>

Imir ifransiwen tekkan ɣef lquyyad akk d Bacaɣat akken ad qehren agdud. Kksen akal i wid yellan zedɣen anda yella wakal yelhan i tyerza, kra snegren-t, kra snejlan-t. Fkan akal nni i Lquyyad. Lquyyad-agi akk d Bacaɣat d wid yeɣran taεrabt, ifransiwen llan lemmden taεrabt akken ad ttemyarun d lquyyad-a, imezdaɣ yeqqimen d aklan deg tmurt nsen yeqqim-asen-d kan ad ḍefren almi i d-ufan iman nsen ttun tutlayt nsen, akken i sen-iruh wakal nsen, ǧǧan iman nsen tuɣ fell-asen taɛrabt am akken werǧin mmeslayen taqbaylit. Ihi akka i teqqurmeḍ tjumma anda tettwameslay teqbaylit ass-a.

[eldin raigmore]

Are these meant to be phonemic transcriptions or romanizations? The use of slashes suggests the former, but there are already standard ways of doing that, and I don't get the sense that's what's being asked here.

Romanizations.
Sorry, I should have written
< th th ht ht > .

And for romanizations, I'd stay away from superscripts and subscripts; makes text kinda ugly and difficult to read if there are a lot of those. Although that may be just my personal opinion.

Well, if he's already in territory where he must use diacritics, I think superscripts and subscripts are better than other kinds of diacritics.
Of course that's just my personal opinion, from anyone else's point of view. (To me, though, it might as well be gospel.)

[eldin raigmore]

... if I had an aspirated voiceless dental plosive, ...

Can people really articulate plosives using the tip or blade of the tongue against the tips of the upper teeth? And have them sound like plosives instead of like somewhat-fortis fricatives?
I can't.
Isn't that what dental means -- the tip of the tongue against the tips or backs of the upper teeth (not the gums nor the alveolar ridge!), or the tips of the upper teeth against the tip or blade of the tongue?

I can see two errors that I might have made that could explain this.
(1) I've misunderstood what "dental" means.
(2) Everyone else's attempt to articulate a dental plosive actually sounds like a plosive, not a fortis fricative; I'm the only one who can't do that.

UPSID Sound Selection gives several languages with dental plosives and several with dental fricatives or dental affricates; but none of those with dental plosives also have dental fricatives or dental affricates.

The IPA doesn't give a way to separate dental from alveolar except among fricatives.

[shmittygoatman]

Can people really articulate plosives using the tip or blade of the tongue against the tips of the upper teeth? And have them sound like plosives instead of like somewhat-fortis fricatives?

That's a good question, and one that I've encountered in some of my conlangs. I personally, if I were to use a dental plosive, would see it like you do; a "somewhat-fortis fricative". My conlang in question, however, differentiates between a voiceless dental plosive /t̪/ and a voiceless dental fricative /θ/.

Create your romanization without worrying too much about making sure that the reader perfectly pronounces the word, but for those with earnest interest, create an appendix in the back that'll help to explain further, like many novels do.

Thanks for the help! I guess I was looking for a system that simply didn't exist. Oh well. At least I'll be able to enjoy all my retroflex plosives, even if the readers won't.