Romanization Challenge Thread v2.0

[bradrn]

I opted for digraph <ng>. I suppose this could be confusing if there's often CVn.gV(C) syllables. For <q> you could use <'>, but I dislike apostrophs-as-letters. <y> is controversial, but seeing you don't need it otherwise and it resembles ɣ, why not? <ll> is taken from Welsh. I've opted not to use it after t, no need to make the spelling 100% phonetic (althoug CVt.lV(C) and CVtl.lV(C) clashes could occur). Vowels are straightforward, schwa can be <e> since there's no /e/, and long vowels can be doubled since syllables are CV(C) and clashes can't occur.

Thanks for replying jal! A couple of comments on the points you make here:

• I don’t much mind clashes with digraphs. (Usually I just add apostrophes to keep them seperate, like in Pinyin).
• I actually really like using ⟨q⟩ for /ʔ/ — I think I’ll end up doing that if I don’t just use ⟨ʔ⟩.
• On the other hand, I don’t much like ⟨y⟩; since you don’t use ⟨g⟩ anywhere, I was thinking of using ⟨g⟩ (the intention was for /ɣ/ to be a lenited */ɡ/).
• I forgot that I could use ⟨ll⟩ for /ɬ/ — looking at your sample text, it’s much more readable than I expected!

/tʰaŋ ˈʔaŋətʰ ˈɬiːsə || bal ˈwiːlət͡s ʔəˈt͡ɬaːɣ || ⁿdil ˈdaːlan ˈt͡suʔəɬ t͡sʰit͡ɬ ˈʔinəɣ/

<Thang qangeth lliise. Bal wiilets qetlaay. Ndil daalan tsuqall tshitl qiney.>

Does the job I think, and pretty readible?

Yes, I think that’s pretty readable. (It’s certainly better than my own attempt at an ASCII-only romanization.)

[Birdlang]

I think you’ll like this other attempt I did at that same one
Consonants:

/m n ŋ/ m n ng
/p t k ʔ/ p t k q
/pʰ tʰ kʰ/ ph th kh
/ᵐb ⁿd ᵑɡ/ mb nd ñg (you could also write it as nh)
/b d/ b d
/ts tɬ/ c j
/tsʰ tɬʰ/ ch jh
/f s ɬ ɣ/ f s x g
/w l j/ v l y

Vowels: /i u ə a iː aː/ i u e a i’ a’

Phonotactics: All syllable have CV(C) structure. The glottal stop is allowed as an onset, but forbidden as a coda. Moras are defined as follows: a CV syllable is 1 mora, a CVː syllable (with a long vowel) is 2 moras, a CVC syllable (with a coda) is also 2 moras, and a CVːC syllable (with both a long vowel and a coda) is 3 moras. Words must be at least bimoraic.

Stress: Stress is predictable: the syllable with the most number of moras (i.e. the heaviest syllable) is stressed, so e.g. /ˈɬii.sə/, /ʔəˈtɬaaɣ/. If there are multiple heaviest syllables, the stress goes on the first, e.g. /ˈba.na/, /daˈlas.naː/. However, a syllable with /ə/ as its vowel may not be stressed; instead, the next heaviest (or next equal heaviest if there are multiple syllables with the same weight) syllable is stressed, so we get e.g. /ˈʔi.nəɣ/ rather than /ʔiˈnəɣ/, /da.ləsˈnaː/ rather than /daˈləs.naː/. Stress is not phonemic, but it may appear to be depending on syllabification, especially with prenasalised consonants: e.g. /wa.ⁿdal/ is stressed on the second syllable, whereas /wan.dal/ is stressed on the first syllable.

Sample text:

/tʰaŋ ˈʔaŋətʰ ˈɬiːsə || bal ˈwiːlət͡s ʔəˈt͡ɬaːɣ || ⁿdil ˈdaːlan ˈt͡suʔəɬ t͡sʰit͡ɬ ˈʔinəɣ/
Thang qangeth xi’se. Bal wi’lec qeja’g. Ndil da’lan cuqexchij qineg.

How do you like this 2nd one I made?

[bradrn]

I think you’ll like this other attempt I did at that same one
Consonants:

/m n ŋ/ m n ng
/p t k ʔ/ p t k q
/pʰ tʰ kʰ/ ph th kh
/ᵐb ⁿd ᵑɡ/ mb nd ñg (you could also write it as nh)
/b d/ b d
/ts tɬ/ c j
/tsʰ tɬʰ/ ch jh
/f s ɬ ɣ/ f s x g
/w l j/ v l y

Vowels: /i u ə a iː aː/ i u e a i’ a’

Phonotactics: All syllable have CV(C) structure. The glottal stop is allowed as an onset, but forbidden as a coda. Moras are defined as follows: a CV syllable is 1 mora, a CVː syllable (with a long vowel) is 2 moras, a CVC syllable (with a coda) is also 2 moras, and a CVːC syllable (with both a long vowel and a coda) is 3 moras. Words must be at least bimoraic.

Stress: Stress is predictable: the syllable with the most number of moras (i.e. the heaviest syllable) is stressed, so e.g. /ˈɬii.sə/, /ʔəˈtɬaaɣ/. If there are multiple heaviest syllables, the stress goes on the first, e.g. /ˈba.na/, /daˈlas.naː/. However, a syllable with /ə/ as its vowel may not be stressed; instead, the next heaviest (or next equal heaviest if there are multiple syllables with the same weight) syllable is stressed, so we get e.g. /ˈʔi.nəɣ/ rather than /ʔiˈnəɣ/, /da.ləsˈnaː/ rather than /daˈləs.naː/. Stress is not phonemic, but it may appear to be depending on syllabification, especially with prenasalised consonants: e.g. /wa.ⁿdal/ is stressed on the second syllable, whereas /wan.dal/ is stressed on the first syllable.

Sample text:

/tʰaŋ ˈʔaŋətʰ ˈɬiːsə || bal ˈwiːlət͡s ʔəˈt͡ɬaːɣ || ⁿdil ˈdaːlan ˈt͡suʔəɬ t͡sʰit͡ɬ ˈʔinəɣ/
Thang qangeth xiise. Bal wiilec qejaag. Ndil daalan cuqexchij qineg.

How do you like this 2nd one I made?

I think I like this one as well! The only thing I don’t like is the use of ⟨j x v⟩ for /tɬ ɬ w/ (I would have used ⟨tl lh w⟩ or something similar), but the rest is nice.

EDIT: I think you have a typo, by the way: in the sample text, you did /ˈwiːlət͡s/ as ⟨wiilec⟩, but I believe that should be ⟨viʼlec⟩.

[Birdlang]

Ok, fixed that.
Here are some of my old refurbished conlangs.
Oreyo
/oˈɍeɉo/
/m n ṉʲ ɴ/
/p b t d ṯʲ ḏʲ k g q ɢ ʔ/
/f v s z ɕ ʑ x ɣ χ ʁ h/
/ts dz ṯʲɕ ḏʲʑ/
/l ḻʲ j ɉ ɥ ɰ w/
/r ɍ ṟʲ/
The r with the bar is a palatal trill and the j with the bar is an alveolopalatal approximant.

/i y ɯ u ɪ ʏ ɯ̽ ʊ e ø ɤ o ə ɛ œ ʌ ɔ a ɑ/
The tense vowels are always long, and the lax ones are always short.

/jus ɴɑm tjɛ.lɪq χə.lɑm ʑi.bɔp qɑ.χɑl || ɣeɍ.hos ṟʲɑ.ri kɥy.lis ɉa.jo kɛṉʲ saj.ha.bə ɢɔɍ ṯʲ͝ɕoɴ.ɢo txi bɑb.ɕo ki.ʔis ||/

Stardot
/βuŋ sʰɑːʀ/
/m n ŋ/
/p b t d k g ʔ/
/β f s sʰ~z̥ ʒ ʝ ɣ χ h/
/l ʎ w/
/r ʀ/

/i iː ʉ ʉː u uː ʏ ʏː ʊ ʊː e eː o oː ɛ ɛː œ œː ɔ ɔː æ æː ɑ ɑː ɒ ɒː/

/koː.tok hɒːʔ χiː.riːm βɑː.soː ʒʀiː.nʎɑːs ŋoːn.tok priʒ ʎɒː.kʏː æːs || hʎʊːŋ ʝʉːs ʀɛs bwiː.ʀiːŋ nʝoːs ʊ.ʎɒːs.ʃʀiːn.nʎʏːs sʰuːg gʀiː.miːs hɑː.toːn ||/

Cotaili
/koːˈtajliː/
/m n ɲ ɲʷ ŋ ŋʷ/
/p b t d c ɟ cʷ ɟʷ k g kʷ gʷ/
/ɸ β θ ð s z ç ʝ çʷ ʝʷ ʃ ʒ ʃʷ ʒʷ x ɣ xʷ ɣʷ h hʷ~ʍ/
/ts dz tʃ dʒ tʃʷ dʒʷ/
/l j ɥ w ɰ/
/r/

/i y ɨ u ə e o œ æ ɑ/
All vowels can be long or short.
Stress is unpredictable and has to be marked.

/iːɲʷ ˈsvuː.hɑːn ˈŋʷeː.nɑː hɑ.ˈliːr ɥœːs ˈhæt.ŋɑ kuː.ˈrim.bus ʃʷriː || kʷɑːs 'kleːn.təs hə.'biːs kwɑːr ʒʷdʒʷoː hœt mɑːns kiːr.'liː ||/

[Pabappa]

This is a new conlang I’m making, but I’m having trouble romanizing it (for the moment I’m just using an IPA-based system). Interestingly, it’s fairly easy to make a romanization — it’s making one that’s fairly intuitive and readable that’s hard. I’d appreciate some help!

My idea is pretty close to your #4 in the last thread, i think, but i didnt see all the consonants so

/m n ŋ/ m n ŋ
/p t k ʔ/ p t k Ø ... i.e. dont bother writing the glottal stop at all, unless it can contrast with hiatus or word-initial vowels
/pʰ tʰ kʰ/ ph th kh
/ᵐb ⁿd ᵑɡ/ mb nd ng ... dont need to bother with writing an ŋ
/b d/ b d
/ts tɬ/ c tl ... i dislike digraphs but /tɬ/ is tricky to cram into one letter
/tsʰ tɬʰ/ ch tlh
/f s ɬ ɣ/ f s ɬ g
/w l j/ w l y ... or maybe use r for /l/ and l for /ɬ/ ... depends how common they each are.

Vowels: /i u ə a iː aː/ i ū u a ī ā, for symmetry's sake, ... helps if there is some grammatical connection between /u/ and /ə/. Alternatively, e u y a i o, which looks neater, though even i would prefer to use macrons here. how much do the long/short vowel pairs differ by quality, if at all?

Oh, I should mention .... its quite odd to have an aspiration distinction without having a phonemic free /h/. Was there an /h/ in this language in the recent past?

yes, i always edit my posts ... i guess if there is no /h/, you could spell the glottal stop as h unless you have clusters like /kʔ/.

[bradrn]

Vowels: /i u ə a iː aː/ i ū u a ī ā, for symmetry's sake, ... helps if there is some grammatical connection between /u/ and /ə/. Alternatively, e u y a i o, which looks neater, though even i would prefer to use macrons here. how much do the long/short vowel pairs differ by quality, if at all?

I wouldn’t know; I’ve just started making this language. For now, just assume there’s no connection between /u/ and /ə/, and there’s no quality contrast between the long and short vowels.

Oh, I should mention .... its quite odd to have an aspiration distinction without having a phonemic free /h/.

This would be because I had no idea aspiration usually implies /h/. Where did you get that from? And it appears to be false anyway: PHOIBLE lists lots of languages with aspiration but no /h/. (Yes, I know PHOIBLE isn’t perfect, but it’s good enough for these sort of searches.)

[bradrn]

Birdlang’s languages:

/oˈɍeɉo/ ⟨Oɍeyro⟩

/m n ṉʲ ɴ/ ⟨m n ny ŋ⟩
/p b t d ṯʲ ḏʲ k g q ɢ ʔ/ ⟨p b t d ty dy k g q gq ʔ⟩
/f v s z ɕ ʑ x ɣ χ ʁ h/ ⟨f v s z sy zy x ɣ xq gq h⟩
/ts dz ṯʲɕ ḏʲʑ/ ⟨ts dz tc dj⟩
/l ḻʲ j ɉ ɥ ɰ w/ ⟨l ly y yr ÿ ẅ w⟩
/r ɍ ṟʲ/ ⟨r ɍ ry⟩

/i y ɯ u/ ⟨i ʉ ɯ u⟩
/ɪ ʏ ɯ̽ ʊ/ ⟨î ʉ̂ ɯ̂ û⟩
/e ø ɤ o/ ⟨e ø ë o⟩
/ə/ ⟨ə⟩
/ɛ œ ʌ ɔ/ ⟨ê ö ʌ ô⟩
/a ɑ/ ⟨a â⟩

/jus ɴɑm tjɛ.lɪq χə.lɑm ʑi.bɔp qɑ.χɑl || ɣeɍ.hos ṟʲɑ.ri kɥy.lis ɉa.jo kɛṉʲ saj.ha.bə ɢɔɍ ṯʲ͝ɕoɴ.ɢo txi bɑb.ɕo ki.ʔis ||/
⟨Yus ŋâm tyêlîq xəlâm zyibôp qâxâl. Ɣeɍhos ryâri kÿʉlis yrayo kêny sayhabə gqôɍ tcoŋgqo txi bâbsyo kiʔis.⟩


/βuŋ sʰɑːʀ/ ⟨Vuꞌŋ zaṟ⟩

/m n ŋ/ ⟨m n ŋ⟩
/p b t d k g ʔ/ ⟨p b t d k g q⟩
/β f s sʰ~z̥ ʒ ʝ ɣ χ h/ ⟨v f s z zh y ɣ x h⟩
/l ʎ w/ ⟨l ly w⟩
/r ʀ/ ⟨r ṟ⟩

/i ʉ u ʏ ʊ e o ɛ œ ɔ æ ɑ ɒ /
⟨i ʉ u ĩ ũ e o ẽ ø õ ã a ɔ⟩

Short vowels marked with saltillo ⟨iꞌ ʉꞌ uꞌ etc.⟩, long vowels unmarked.

/koː.tok hɒːʔ χiː.riːm βɑː.soː ʒʀiː.nʎɑːs ŋoːn.tok priʒ ʎɒː.kʏː æːs || hʎʊːŋ ʝʉːs ʀɛs bwiː.ʀiːŋ nʝoːs ʊ.ʎɒːs.ʃʀiːn.nʎʏːs sʰuːg gʀiː.miːs hɑː.toːn ||/
⟨Kotoꞌk hɔq xirim vaso zhṟinlyas ŋontoꞌk priꞌzh lyɔkĩ ãs. Hlyũŋ yʉs ṟẽꞌs bwiṟiŋ nyos ũꞌlyɔsshṟinnlyĩs zug gṟimis haton.⟩



/koːˈtajliː/ ⟨Kotáꞌyli⟩

/m n ɲ ɲʷ ŋ ŋʷ/ ⟨m n ɲ ɲw ŋ ŋw⟩
/p b t d c ɟ cʷ ɟʷ k g kʷ gʷ/ ⟨p b t d c j cw jw k g kw gw⟩
/ɸ β θ ð s z ç ʝ çʷ ʝʷ ʃ ʒ ʃʷ ʒʷ x ɣ xʷ ɣʷ h hʷ~ʍ/ ⟨f v ? ? s z ch jh chw jhw sh zh shw zhw x ɣ xw ɣw h hw⟩
/ts dz tʃ dʒ tʃʷ dʒʷ/ ⟨ts dz tc dj tcw djw⟩
/l j ɥ w ɰ/ ⟨l y ÿ w ẅ⟩
/r/ ⟨r⟩

/i y ɨ u ə e o œ æ ɑ/ ⟨i ʉ ɨ u ə e o ö ä a⟩

Short vowels marked with saltillo ⟨iꞌ ʉꞌ ɨꞌ etc.⟩, long vowels unmarked.
Stress is marked with acute accept ⟨í ʉ́ ɨ́ etc.⟩, except for ⟨ö ä⟩, which become ⟨ő a̋⟩ when stressed.

/iːɲʷ ˈsvuː.hɑːn ˈŋʷeː.nɑː hɑ.ˈliːr ɥœːs ˈhæt.ŋɑ kuː.ˈrim.bus ʃʷriː || kʷɑːs 'kleːn.təs hə.'biːs kwɑːr ʒʷdʒʷoː hœt mɑːns kiːr.'liː ||/
⟨Iɲw svúhan ŋwéna haꞌlír ÿös ha̋ꞌtŋaꞌ kuríꞌmbuꞌs shwri. Kwas kléntəꞌs həꞌbís kwar zhwdjwo höt mans kirlí.⟩

[jal]

On the other hand, I don’t much like ⟨y⟩; since you don’t use ⟨g⟩ anywhere, I was thinking of using ⟨g⟩

You are right! I thought your language had /g/, and I used it, but since I didn't, <g> is a way better choice :). I also just saw I used <ng> for both /ŋ/ and /ⁿg/, but if you don't have initial /ŋ/ that shouldn't pose a problem. Anyway, glad I could give you some ideas :).


JAL

[bradrn]

I also just saw I used <ng> for both /ŋ/ and /ⁿg/, but if you don't have initial /ŋ/ that shouldn't pose a problem.

Well, I do happen to have initial /ŋ/, so that might be a problem.

Anyway, glad I could give you some ideas .

It was helpful, so thank you!

[jal]

Well, I do happen to have initial /ŋ/, so that might be a problem.

Well, you could always go for <ngg>.


JAL

[Birdlang]

Ok figured out a romanization for my 3 Langs using 1 character per sound in most cases
Oreyo
/oˈɍeɉo/ Ořeǰo
/m n ṉʲ ɴ/ m n ň ŋ
/p b t d ṯʲ ḏʲ k g q ɢ ʔ/ p b t d ť ď k g q ġ ʔ
/f v s z ɕ ʑ x ɣ χ ʁ h/ f v s z š ž x ḡ ḫ ǥ h
/ts dz ṯʲɕ ḏʲʑ/ c ʒ č ǯ
/l ḻʲ j ɉ ɥ ɰ w/ l ľ j ǰ w̌ y̌ w
/r ɍ ṟʲ/ r ř ŗ
The r with the bar is a palatal trill and the j with the bar is an alveolopalatal approximant.

/i y ɯ u ɪ ʏ ɯ̽ ʊ e ø ɤ o ə ɛ œ ʌ ɔ a ɑ/ í û î ú i ü ï u é ô ê ó ǝ e ö ë o a á
The tense vowels are always long, and the lax ones are always short.

/jus ɴɑm tjɛ.lɪq χə.lɑm ʑi.bɔp qɑ.χɑl || ɣeɍ.hos ṟʲɑ.ri kɥy.lis ɉa.jo kɛṉʲ saj.ha.bə ɢɔɍ ṯʲ͝ɕoɴ.ɢo txi bɑb.ɕo ki.ʔis ||/

Stardot
/βuŋ sʰɑːʀ/ Vuñ Zāq
/m n ŋ/ m n ñ
/p b t d k g ʔ/ p b t d k g ʼ
/β f s sʰ~z̥ ʒ ʝ ɣ χ h/ v f s z ž j ģ x h
/l ʎ w/ l y w
/r ʀ/ r q

/i iː ʉ ʉː u uː ʏ ʏː ʊ ʊː e eː o oː ɛ ɛː œ œː ɔ ɔː æ æː ɑ ɑː ɒ ɒː/ i ī ú û u ū ü ǖ ʊ ʋ é ê ó ô e ē ö ȫ o ō æ ǣ a ā ä ǟ

/koː.tok hɒːʔ χiː.riːm βɑː.soː ʒʀiː.nʎɑːs ŋoːn.tok priʒ ʎɒː.kʏː æːs || hʎʊːŋ ʝʉːs ʀɛs bwiː.ʀiːŋ nʝoːs ʊ.ʎɒːs.ʃʀiːn.nʎʏːs sʰuːg gʀiː.miːs hɑː.toːn ||/

Cotaili
/koːˈtajliː/ Kōtáilī
/m n ɲ ɲʷ ŋ ŋʷ/ m n ñ ñw ŋ ŋw
/p b t d c ɟ cʷ ɟʷ k g kʷ gʷ/ p b t d ǩ ǧ ǩw ǧw k g kw gw
/ɸ β θ ð s z ç ʝ çʷ ʝʷ ʃ ʒ ʃʷ ʒʷ x ɣ xʷ ɣʷ h hʷ~ʍ/ f v þ ð s z ś ź św źw š ž šw žw x ǥ xw ǥw h hw
/ts dz tʃ dʒ tʃʷ dʒʷ/ c dz č dž čw džw
/l j ɥ w ɰ/ l j/i ÿ/ü w/u ẅ/û
/r/ r

/i y ɨ u ə e o œ æ ɑ/ i ü y u ä e o ö æ a
All vowels can be long or short. Long marked with macron.
Stress is unpredictable and has to be marked. Marked with acute accent.

/iːɲʷ ˈsvuː.hɑːn ˈŋʷeː.nɑː hɑ.ˈliːr ɥœːs ˈhæt.ŋɑ kuː.ˈrim.bus ʃʷriː || kʷɑːs 'kleːn.təs hə.'biːs kwɑːr ʒʷdʒʷoː hœt mɑːns kiːr.'liː ||/

[Nortaneous]

here is a very cursed thing

Ssob bbaek vaaf maht jeif gaij niaao sem ai mengl hen. Jong luk ddox, mout ssu kehk mak moot aaox kuaahk eemx ddi gua. Hol ssaao geof vaeix niaao ssangf, ai laf ngeaix ddaew dien maaf gox teiv bbaht ngeaix ssu kehk gie, goeiv aaox, tif sej, ddeak goi lax mengl hoen. Sao ddox ssox bbaehk bbaht ddaeof touiv mengl heiht seehk gox, nieeng gangf touiv, tif ngeaif ssu kehk gaangf sseej jat aaox kuaahk, vaf guix gungf, ssox bbaehk ddaaj faaiv deef bbov.

[bradrn]

here is a very cursed thing

Ssob bbaek vaaf maht jeif gaij niaao sem ai mengl hen. Jong luk ddox, mout ssu kehk mak moot aaox kuaahk eemx ddi gua. Hol ssaao geof vaeix niaao ssangf, ai laf ngeaix ddaew dien maaf gox teiv bbaht ngeaix ssu kehk gie, goeiv aaox, tif sej, ddeak goi lax mengl hoen. Sao ddox ssox bbaehk bbaht ddaeof touiv mengl heiht seehk gox, nieeng gangf touiv, tif ngeaif ssu kehk gaangf sseej jat aaox kuaahk, vaf guix gungf, ssox bbaehk ddaaj faaiv deef bbov.

Oddly enough, I actually quite like this one. (Maybe it’s because the tone letters seem much more obvious than usual.)

[bradrn]

Recently I published a sketch of a conlang with a very interesting phonology. I’d be interested to see if anyone else can come up with a romanization of it! (Well, I already did, but I’d like to see what everyone else can come up with.)

The phoneme inventory is:

	Labial	Coronal	Palatal	Rounded Palatal	Velar	Roudned velar	Laryngeal	Rounded laryngeal
Occlusive	b~m̩	d~n̩	ɟ~ɲ̍	ɟʷ~ɲ̍ʷ	ɡ~ŋ̍	ɡʷ~ŋ̍ʷ
Unvoiced continuant	f~ʋ̤̍	ʃ~ɚ̤	ç~i̤	çʷ~ÿ	x~ɯ̤	xʷ~ṳ	ħ~ɑ̤	ħʷ~ɔ̤
Voiced continuant	v~ʋ̩	ʒ~ɚ	ʝ~i	ʝʷ~y	ɣ~ɯ	ɣʷ~u	ʕ~ɑ	ʕʷ~ɔ
Glottalised	ɓ~ʋ̰̍	ɗ~ɚ̰	ʄ~ḭ	ʄʷ~y̰	ɠ~ɯ̰	ɠʷ~ṵ	ʔ~ɑ̰	ʔʷ~ɔ̰

As shown on the table, each phoneme has both consonantal and syllabic realisations. For words of more than one phoneme, the consonantal realisation is used for the first phoneme; subsequent phonemes alternate between syllabic and consonantal realisations.

Sample sentences:

/ʕvbɟʷɣgɣʷʝ bʕʷdɓʔɣʷʝ bʒdɠʄdʕx/
[ʕʋ̩bɲ̍ʷɣŋ̍ɣʷi bɔdʋ̰̍ʔuʝ bɚdɯ̰ʄn̩ʕɯ̤]

/ɓʔɣʷʝ bʒɣʷʝ ʕɣʷʄħ dbʒɠʄʝʝʔ/
[ɓɑɣʷi bɚɣʷi ʕuʄɑ̤ dm̩ʒɯ̰ʄiʝɑ]

/bʕʷdbʒɣʷʝ dɣʷʄħbʒɣʷʝ dɠʄdɠʄvħʷ/
[bɔdm̩ʒuʝ duʄɑ̤bɚɣʷi dɯ̰ʄn̩ɠḭvɔ̤]

[jal]

Recently I published a sketch of a conlang with a very interesting phonology. I’d be interested to see if anyone else can come up with a romanization of it! (Well, I already did, but I’d like to see what everyone else can come up with.)

Though I'm not going to burn my fingers on this one :), I would probably expect a phonemic writing system, if only because at least it looks sane on paper, if not in pronunciation :).


JAL

[bradrn]

I would probably expect a phonemic writing system, if only because at least it looks sane on paper, if not in pronunciation .

I must admit that I’m surprised to hear this. Do you seriously think that /dɣʷʄħbʒɣʷʝ/ looks more sane than [duʄɑ̤bɚɣʷi]‽

[jal]

I must admit that I’m surprised to hear this. Do you seriously think that /dɣʷʄħbʒɣʷʝ/ looks more sane than [duʄɑ̤bɚɣʷi]‽

In a romanization (or other script), the root of a word would readibly be visible, even though the surface pronunciation would be different. For example, take your examples from the other thread:

house /bʕʷd/ bod → /d-bʕʷd/ dmḧn
tree /ɣʷʄħ/ ɣ̈ĩħ → /d-ɣʷʄħ/ duɉä
sun /vbɟʷɣg/ vmj̈ɯg → /d-vbɟʷɣg/ dwbñʷɣŋ

Now let’s add a definite affix /ʕ-/:

/ʕ-bʕʷd/ hmḧn, /ʕ-d-bʕʷd/ hnbod
/ʕ-ɣʷʄħ/ huɉä, /ʕ-d-ɣʷʄħ/ hnɣ̈ĩħ
/ʕ-vbɟʷɣg/ hwbñʷɣŋ, /ʕ-vbɟʷɣg/ hnvmj̈ɯg

So we cycle, with a phonetic transcription, through:

• <bod>, <dmḧn>, <hnbod>
• <ɣ̈ĩħ>, <duɉä>, <hnɣ̈ĩħ>
• <vmj̈ɯg>, <dwbñʷɣŋ>, <hnvmj̈ɯg>

However, with a phonemic transcription, this could be:

• <bod>, <dbod>, <hnbod>
• <gih>, <dhig>, <hngih>
• <vmjug>, <dvmjug>, <hnvmjug>

Which makes it a lot clearer that there's a root word with affixes, even though the surface forms are quite different.


JAL

[bradrn]

So we cycle, with a phonetic transcription, through:

• <bod>, <dmḧn>, <hnbod>
• <ɣ̈ĩħ>, <duɉä>, <hnɣ̈ĩħ>
• <vmj̈ɯg>, <dwbñʷɣŋ>, <hnvmj̈ɯg>

However, with a phonemic transcription, this could be:

• <bod>, <dbod>, <hnbod>
• <gih>, <dhig>, <hngih>
• <vmjug>, <dvmjug>, <hnvmjug>

Which makes it a lot clearer that there's a root word with affixes, even though the surface forms are quite different.

Ah, right. But in that case, how do you decide which phonemes to write as consonants and which to write as vowels?

[Richard W]

This would be because I had no idea aspiration usually implies /h/. Where did you get that from? And it appears to be false anyway: PHOIBLE lists lots of languages with aspiration but no /h/. (Yes, I know PHOIBLE isn’t perfect, but it’s good enough for these sort of searches.)

It's contrastive aspiration that is supposed to require /h/. It's one of the arguments used when trying to reconstruct the PIE consonant system. However, I see that [x] (or backer) seems to stand in well enough for that rule. [ɦ] seems to be another adequate form - see post-Sanskrit Indic. Ionian Greek seems to have managed to eliminate /h/; but in fairly short order, the aspirated voiceless stops then fricativised. (It's quite common for voiceless aspirates to have fricative allophones.)

[bradrn]

This would be because I had no idea aspiration usually implies /h/. Where did you get that from? And it appears to be false anyway: PHOIBLE lists lots of languages with aspiration but no /h/. (Yes, I know PHOIBLE isn’t perfect, but it’s good enough for these sort of searches.)

It's contrastive aspiration that is supposed to require /h/. It's one of the arguments used when trying to reconstruct the PIE consonant system. However, I see that [x] (or backer) seems to stand in well enough for that rule. [ɦ] seems to be another adequate form - see post-Sanskrit Indic. Ionian Greek seems to have managed to eliminate /h/; but in fairly short order, the aspirated voiceless stops then fricativised. (It's quite common for voiceless aspirates to have fricative allophones.)

Ah, interesting. Guess I’ll have to revisit the phonology then — after I’ve already done four months’ worth of work on the rest of the language! (I wonder if it’s sufficient to have /h/ be really uncommon…)

EDIT: But once again, PHOIBLE gives some counterexamples. (Interestingly, quite a lot of those seem to have /ɣ/ — if the presence of /ɣ/ is sufficient to allow contrastive aspiration, then that would be really convenient for me, as I already have /ɣ/ in my phoneme inventory.)