Romanization Challenge Thread v2.0

[bradrn]

Thanks to the discussion on another thread, I’ve finally managed to find a romanisation of Mandarin Chinese I’m happy with:

/m n/ ⟨m n⟩
/p t k/ ⟨b d g⟩
/pʰ tʰ kʰ/ ⟨p t k⟩
/ts tʂ tɕ/ ⟨dz j j⟩
/tsʰ tʂʰ tɕʰ/ ⟨ts ch ch⟩
/f s ʂ ɕ x/ ⟨f s sh sh h⟩
/w l ɹ j/ ⟨w l r y⟩

[ɹ̩ əɹ] ⟨r er⟩
[ɤ ɐ ei ae oʊ ɑo] ⟨e a ei ai ou ao⟩
[ən an ɐŋ ɑŋ] ⟨en an ang aang⟩
[i jɛ jɐ joʊ jɑʊ] ⟨i ye ya you yau⟩
[in iɛn iŋ iɑŋ] ⟨in yen ing yang⟩
[u wo wɐ weɪ wae] ⟨u wo wa wei wai⟩
[wən wan ʊŋ wɑŋ] ⟨wen wan ung wang⟩
[y yɛ yɪn yan yʊŋ] ⟨ü üe üen üan yung⟩

/ma55 ma35 ma21 ma51 ma/ ⟨mā má ma̱ mà ma⟩

Sample text (taken from Omniglot):

人人生而自由,在尊严和权利上一律平等。他们赋有理性和良心,并应以兄弟关系的精神互相对待。

Rénrén shāng ér dzr̀yóu, dzài dzwēnyán hé chüánlì shàang īlǜ píngda̱ng. Tāmen fùyo̱u li̱shìng hé lyángshīn, bìng īng i̱ shyūngdì gwānshì de jīngshén hùshyāng dwèidài.

Compare to Pinyin:

Rénrén shēng ér zìyóu, zài zūnyán hé quánlì shàng yīlǜ píngděng. Tāmen fùyǒu lǐxìng hé liángxīn, bìng yīng yǐ xiōngdì guānxì de jīngshén hùxiāng duìdài.

[Rounin Ryuuji]

I would call that certainly an improvement on Pinyin.

[Travis B.]

I do like that romanization of Mandarin!

[Nortaneous]

Thanks to the discussion on another thread, I’ve finally managed to find a romanisation of Mandarin Chinese I’m happy with:

/m n/ ⟨m n⟩
/p t k/ ⟨b d g⟩
/pʰ tʰ kʰ/ ⟨p t k⟩
/ts tʂ tɕ/ ⟨dz j j⟩
/tsʰ tʂʰ tɕʰ/ ⟨ts ch ch⟩
/f s ʂ ɕ x/ ⟨f s sh sh h⟩
/w l ɹ j/ ⟨w l r y⟩

[ɹ̩ əɹ] ⟨r er⟩
[ɤ ɐ ei ae oʊ ɑo] ⟨e a ei ai ou ao⟩
[ən an ɐŋ ɑŋ] ⟨en an ang aang⟩
[i jɛ jɐ joʊ jɑʊ] ⟨i ye ya you yau⟩
[in iɛn iŋ iɑŋ] ⟨in yen ing yang⟩
[u wo wɐ weɪ wae] ⟨u wo wa wei wai⟩
[wən wan ʊŋ wɑŋ] ⟨wen wan ung wang⟩
[y yɛ yɪn yan yʊŋ] ⟨ü üe üen üan yung⟩

/ma55 ma35 ma21 ma51 ma/ ⟨mā má ma̱ mà ma⟩

Differing from Yale in <ao ang aang yen ü> and replacing the caron with an underline?

(seems reasonable, <ao> is the iconic Mandarin romanization digraph)

[bradrn]

Differing from Yale in <ao ang aang yen ü> and replacing the caron with an underline?

Huh, is it really that close to Yale? I didn’t realise. In any case, these changes are justified: ⟨eng ang⟩ for /ɐŋ ɑŋ/ is confusing, ⟨yan⟩ for /jɛn/ is even more so, ⟨yw yu⟩ are non-obvious for /y/, the caron obscures the tone’s phonemic status as a fairly simple low tone, and I distinctly recall Kuchigakatai saying that the ⟨ao⟩ rhyme does indeed sound like [ao] to them.

[Emily]

i don't speak a word of mandarin but it's puzzling to me that there's two phonemes each covered by <j> <ch> <sh>. shouldn't a romanization try to avoid these ambiguities?

[Rounin Ryuuji]

They're actually context-dependent allophones, note the palatals being represented in practice as "chy jy" and the retroflexes as "ch j".

[bradrn]

i don't speak a word of mandarin but it's puzzling to me that there's two phonemes each covered by <j> <ch> <sh>. shouldn't a romanization try to avoid these ambiguities?

Rounin Ryuuji is correct. These are in complementary distribution: /tɕ tɕʰ ɕ/ occur only before /i y j ɥ/, and /tʂ tʂʰ ʂ k kʰ x/ occur everywhere else. The only reason I showed them separately is that some transcription systems write them differently (e.g. Pinyin), so I wanted to make it clear that I was merging them.

[Emily]

gotcha

[Travis B.]

i don't speak a word of mandarin but it's puzzling to me that there's two phonemes each covered by <j> <ch> <sh>. shouldn't a romanization try to avoid these ambiguities?

Rounin Ryuuji is correct. These are in complementary distribution: /tɕ tɕʰ ɕ/ occur only before /i y j ɥ/, and /tʂ tʂʰ ʂ k kʰ x/ occur everywhere else. The only reason I showed them separately is that some transcription systems write them differently (e.g. Pinyin), so I wanted to make it clear that I was merging them.

If anything, the reason for marking them distinctly is that they neither belong with solely the /tʂ tʂʰ ʂ/ set nor with the /k kʰ x/ set, due to involving a merger before /i y j ɥ/ of the two sets in the recent past (recent enough to explain names such as "Peking" in (dated) English).

[Nortaneous]

Differing from Yale in <ao ang aang yen ü> and replacing the caron with an underline?

Huh, is it really that close to Yale? I didn’t realise.

I guess there's only so many ways to romanize Mandarin in accordance with English loan-stratum orthographic conventions.

But what about the conventions of Zzyxwqnp?

/pʰ tʰ kʰ/ <p t k>
/p t k/ <b d g>
/tsʰ tʂʰ tɕʰ/ <c rh ch>
/ts tʂ tɕ/ <z dr ch>
/f s ʂ ʐ ɕ x/ <f s sh r x h>
/l/ <l>

/0 ɤ ei ou ən əŋ a ai au an aŋ/ <y z e o zn zg a ai au xn ag>
/i je jou in iŋ ja jau jɛn jaŋ/ <i ie io in ig x iau ixn iag>
/u wo wei wən ʊŋ wa wai wan waŋ/ <u uz ue un ug q uqi uqn uqg>
/y ɥe yn jɛŋ ɥen/ <v ve vn vg ven vxn>

(with excrescent -w -y -w -y in the absence of an initial)

<˥ ˧˥ ˨˩˦ ˥˩ 0> <-t -x -p -f 0>

txntshzgt wqnfwuf yip yxgp rznx
rznx wux yitshxnf yip bauf txnt
shat shat shat shat shat shat shat

[bradrn]

i don't speak a word of mandarin but it's puzzling to me that there's two phonemes each covered by <j> <ch> <sh>. shouldn't a romanization try to avoid these ambiguities?

Rounin Ryuuji is correct. These are in complementary distribution: /tɕ tɕʰ ɕ/ occur only before /i y j ɥ/, and /tʂ tʂʰ ʂ k kʰ x/ occur everywhere else. The only reason I showed them separately is that some transcription systems write them differently (e.g. Pinyin), so I wanted to make it clear that I was merging them.

If anything, the reason for marking them distinctly is that they neither belong with solely the /tʂ tʂʰ ʂ/ set nor with the /k kʰ x/ set, due to involving a merger before /i y j ɥ/ of the two sets in the recent past (recent enough to explain names such as "Peking" in (dated) English).

Yes… and this is complicated by the fact that IIRC native-speaker intuition groups the palatals with the velar series, not the retroflex one. But I was making a romanisation primarily for English-speakers, not Mandarin-speakers.

Differing from Yale in <ao ang aang yen ü> and replacing the caron with an underline?

Huh, is it really that close to Yale? I didn’t realise.

I guess there's only so many ways to romanize Mandarin in accordance with English loan-stratum orthographic conventions.

Yep, pretty much. In fact my main issue with Pinyin is that it doesn’t consistently follow those conventions. (Or any others I can figure out, for that matter.)

But what about the conventions of Zzyxwqnp?

Horrible!

…incidentally, you don’t happen to have a reference of these conventions, do you? I’ve been wondering about them for quite some time.

[Nortaneous]

But what about the conventions of Zzyxwqnp?

Horrible!

…incidentally, you don’t happen to have a reference of these conventions, do you? I’ve been wondering about them for quite some time.

/p b (pɕ bʑ) t d ts dz tʂ dʐ tɕ dʑ k g (q ʁ)/ <p b pi bi t d c z rh dr ch j k g kh gh>
/f (fɕ θ) s ʂ ɕ h (χ)/ <f fi hl s sh x h hh>
/m (mȵ) n̪ n ȵ ŋ/ <m mni n hn ny ng>
/mb (mbʑ) nd ndz ndʑ ŋg/ <bb bbi dd zz jj gg>
/w (z) r j/ <w ss r y>

/a ɒ e ə o i u z̩ v̩/ <a q e z o i u y v> + nasality <-n>
/˥ ˧ ˩ ˥˩/ <-t -l -p -x>

/ɡo˧ ru˩dʑə dʑo˥ fɒ˩ pɕə˩ɕi˧ɕi˧ wɒ˥dʑə hu˩ ke˩ tsa˥tsz̩˩je m̩˩/
gol rupjzn jot fqp pizpxilxil wqtjzn hup kep catcypye mp

Although for distributional reasons the vowel system is more like /ja a wa jə ə wə ji wu i u/ <ia~a a q e~z z o i u y v>, and the morpheme /-jə/ <ye> might be an exception to the rule that <z> is written instead of <e> when palatality is indicated on the consonant. (Palatality dissimilation is common in Vengic; cf. Hlu *-it > -ɯc etc.)

[Zju]

/a ɒ e ə o i u z̩ v̩/ <a q e z o i u y v>

/ə z̩/ <z y>

Why? Just... why?

[Rounin Ryuuji]

Orthographic trolling, I would guess.

[Nortaneous]

/a ɒ e ə o i u z̩ v̩/ <a q e z o i u y v>

/ə z̩/ <z y>

Why? Just... why?

<y> as in Nuosu and <z> as in Natqgu, and /ɒ ə/ were later additions. Also, /z̩ v̩/ become something like [zɨ vɨ] or [ɨ wɨ] in many dialects.

[Nortaneous]

Anyway what if Rau was also like that

Initials:
/p b t d dʐ~r kʲ ɡʲ k ʕ/ <p b t d dr k gi q g>
/θ ð s̺ z̺ ɬ ɮ xʲ ɣʲ x ɣ/ <hs h hr r hl l hy y hx x>
/m n ŋ/ <m n ñ>

Finals:
/a e ə o i ɯ/ <a e v o i w>
/au jəu əu jɯu ɯu/ <au eu ou iu u>
/æ ø y/ <c co cu>
/aɹ eɹ əɹ oɹ jʌɹ/ <ar er vr or ir>
/ai ei əi oi ɯi/ <ci ei vi oi wi>
/aɣ əɣ/ <aw vw>
/ɑˤ iɑˤ ɔˤ iʌˤ uoˤ/ <ag eg og ig ug>
/-θ -s̺ -ɬ -xʲ -x/ <s r l y x>

Tones:
/55 21 15/ <-f -j -z>

/m.broiθ.tʰə mnɑ̀ˤ β.ɬqʰə.dɯ̀ɣ.nə xa.ʕjɯ́u xɕʰiabɑ̀ˤ tʰɯ.z̺.β.s̺æ̀/
mbrôisztv mnagj blqvdwgjnv xagiuf xhsebagj twrbhscj

(vs. in bzỏgs te mnàr blqudoù na harib hshtir boà tou ash beshtád)

[foxcatdog]

*p *t *k *kʷ
*s *sʲ
*r *l
*j *w
*m *n *nʲ

*i *u
*ɛ *ɔ
*a

*ɛ and *ɔ generally did not occur in non-initial syllables except in loanwords interdialectal borrowings and compounding
*j did not occur initially except in loanwords
Generally dialects add *ts or lose *r
*p and *k were fricated intervocalically
*m was lenited to a nasalised approximate intervocalically
*j was *ð initially before *o and *u when it did occur there
*mp, *nt and *nk when they did occur had the stop element voiced
2 identical vowels in sequence or *ai, *au and sometimes *oi generally form diphtongs

CVC syllable structure maximally with stops not being able to occur before another consonant and Nasal + Stop sequences being only found in borrowings

Sample words
*kili "fish", *ina "dog", *sulin "river", *sari "energy", *naitɛ "needle", *mɛmakɛ "son", *tasisi "daughter", *mɛma "mother", *taja "father", *naai "chicken", *iia "ladle", *inɔ "fox", *siwimɔ "dragon", *ɛsi "thunder", *pakʷa "axe", *isʲi "elf", *sai "boy", *miata "moon", *umi "sea", *nʲarra "family", sʲuu "a type of song", *kʷɔna "puppy"

[bradrn]

/p t k kʷ/ ⟨p t k kw⟩
/s sʲ (ts)/ ⟨s sy (c)⟩
/r l/ ⟨r l⟩
/j w/ ⟨y w⟩
/m n nʲ/ ⟨m n ny⟩

/i u ɛ ɔ a/ ⟨i u e o a⟩
/ii uu ɛɛ ɔɔ aa ai au oi/ ⟨ii uu ee oo aa ai au oi⟩

/kili ina sulin sari naitɛ/ ⟨kili ina sulin sari naite⟩
/mɛmakɛ tasisi mɛma taja naai/ ⟨memake tasisi mema taya naai⟩
/iia inɔ siwimɔ ɛsi pakʷa/ ⟨iiia ino siwimo esi pakwa⟩
/isʲi sai miata umi nʲarra sʲuu kʷɔna/ ⟨isyi sai miata nyarra syuu kwona⟩

[Darren]

Something inspired by Mbato:

/ɠ̥ɵ́kʰlɪ̃̀ɟɵ̀ʛwʉ̀/

/b͡pʰ tʰ ɟ͡cʰ kʰ/
/t ɟ k q/
/ɓ ɗ̥ ʄ ɠ̥ ʛ/
/ɓ~w~ŋ͡m~ŋ͡ʘ ɗ~l~n ʄ~j~ɲ/
/f s x/

+ATR /i ʉ u e ɵ/
–ATR /ɪ ʊ ɔ ɵ a/
nasal /ɪ̃ ʊ̟̃ ɞ̃/

The protolang had two stop series *T D and an approximant series *R. The *D series became aspirated and the *R series hardened into stops, leaving a PIE-style *T Dʰ D; this was then reinterpreted as a three-way explosiveness contrast, realised as Tʰ T Ɗ̥. Voicing is a secondary feature, with the bilabial and palatal series being voiced and everything else voiceless (other than the uvular implosive) regardless of MOA. /b͡pʰ ɟ͡cʰ/ are true aspirated voiced stops; voicing occurs in the initial closure, followed by a breif period of voicelessness and then the resumption of voicing during the following vowel/approximant. Phrase-initially they're voiceless /pʰ cʰ/.
The fourth row is a sonorant series, which are realised as implosives word-initially (merging with /ɓ/ and /ʄ/, but not /ɗ̥/), as approximants [w l j] word-internally and as nasals [ŋ͡m n ɲ] before a nasal vowel. The labiovelar nasal alternates with a nasal bilabial click [ŋ͡ʘ] word-initially. These are given as /w l j/ in phonemic representation to prevent confusion with the nonexplosive stops.
Syllable structure is (C)(w,l,j)V. There are three tones /ì í î/, between zero and two of which can be attached to a syllable. A zero-tone syllable takes L tone after L or HL, and H after H or phrase-initially. Two tones can only appear on syllables at the right edge of a phrase, where they are generated due to floating tones, which crop up in morphology and push everything rightwards as much as possible within a phrase (a unit indicated in the sample through words connected with an equals sign).

Sample:
/wékwɵ́ ʛà=fwɪ̃̀=kɪ̃wê wakâkʰâ. wɞ̃̀wé=lɵ=^ ʄɵwʉ̀lwɵ̀ tɵ=´=lè=tʰʊ̟̃̀tlù tɵ=`=wʊ̟̃̀lwû. slɪ́ʛɪ lù=fwê ɵ́ê=`=fɞ̃̂. fɪ̃ʄʉb͡pʰjʉ̂/
[ɓe˥kwɵ˥ ʛa˥ fŋ͡mɪ̃˩ kɪ̃˩we˥˩ ɓa˩ka˥˩kʰa˥˩ ǁ ŋ͡mɞ̃˩we˥lɵ˥˩ ʄɵ˩wʉ˩lwɵ˩ tɵ˥ le˩ tʰʊ̟̃˩tlu˩ tɵ˩ ŋ͡mʊ̟̃˩lwu˥˩ ǁ slɪ˥ʛɪ˥ ɗu˩ fwe˥˩ ɵ˥e˥˩ fɞ̃˩˥˩ ǁ fɪ̃˥ʄʉ˥b͡pʰjʉ˥˩]