From X-SAMPA phonology to screenreader-friendly script

[finlay]

o-horn and u-horn are /M/ and /7/ are they not? But actually I think they're the central vowels.

Also /f/ and /k/ are usually written ph and c, for historical reasons because the orthography was derived from Portuguese in the 18th or 19th century.

[Trebor]

o-horn and u-horn are /M/ and /7/ are they not? But actually I think they're the central vowels.

So, /1/, /@/, and /@:/ are misrepresentations of the true phonemic inventory of Vietnamese? Or is it the case that [1, @, @:] are in free variation with [M, 7, 7:], respectively?

(I understand that Japanese is the Asian language you've been working on, but surely we have someone here that speaks or studies this crazy six-tone tongue...)

Also /f/ and /k/ are usually written ph and c, for historical reasons because the orthography was derived from Portuguese in the 18th or 19th century.

I knew that some Portuguese individuals originally designed a Latin script for Vietnamese, so I understood how <ph> and <c> came to be used for /f/ and /k/, but I don't see any reason for maintaining this arrangement. I do find <nh> an elegant solution for /J/, even though it's Portuguese-influenced.

[finlay]

We used to have a Vietnamese member once upon a time, they disappeared though. I can't remember their username. I think there's at least one person who can use it.

Anyway, I guess I'm a little confused as to your goal – because if you actually want to learn Vietnamese and/or communicate in Vietnamese, I'd recommend one of the ASCII options that they generally use to input text, and I'd recommend not trying to change too much, as there might be some passing familiarity with those options among Vietnamese speakers. But if it's just a personal pet project for you to visualize (as it were) the phonology, go ahead.

[Nortaneous]

/f x/ are written <ph kh> because they used to be /p_h k_h/ -- and apparently <kh> is still /k_h/ in some dialects. <th> is still /t_h/ though.

p t ts` c k <p t tr ch k>
t_h k_h~x <th kh>
b_< d_< <b d>
f v s z s` G h <f v x z s g h>
m n J N <m n nh ng>
l r j w <l r i u>

a a: E @ O e @: o i 1 u <a aa e y o ee yy oo i w u>
aj a:j @j @:j Oj oj 1j uj <ai aai yi yyi oi ooi wi ui>
aw a:w @w Ew ew 1w iw <au aau yu eu eeu wu iu>
wi <ui>
i@ 1@ u@ <ie ye ue>
i@w u@j 1@w 1@j <ieu uei weu wei>

I'm not sure what diacritics screenreaders can read, so I'll use tone letters.
ngang huyen sac nang hoi nga = <0 j z q v x>
mid ˨˩ ˧˥ ˧ˀ˨ʔ ˧˩˧ ˧ˀ˥

IPA: tʰoj˧ˌ cuŋ˧˥ ta˧ haj˧ˀ˥ʔ suəŋ˧˥ˌ lam˨˩ lon˧ˀ˨ʔ son˧ˀ˨ʔ tieŋ˧˥ nɔj˧˨ kuə˧˩˧ cuŋ˧˥ nɔ˧˥ cɔ˧ hɔ˧ˀ˨ʔ ŋɛ˧ kʰoŋ˧ dɯək˧ˀ˨ʔ tiəŋ˧˥ nɔi˧˥ kuə˧˩˧ ŋɯəj˨˩ naːj˨˩ vəːj˧˥ ŋɯəj˨˩ kiə˧
X-SAMPA: t_hoj_M" cuN_M_T ta_M haj_M_>_T? su@N_M_T" lam_L_B lon_M_>_L? son_M_>_L? tieN_M_T nOj_M_L ku@_M_B_M cuN_M_T nO_M_T cO_M hO_M_>_L? NE_M k_hoN_M dM@k_M_>_L? ti@N_M_T nOi_M_T ku@_M_B_M NM@j_L_B na:j_L_B v@:j_M_T NM@j_L_B ki@_M

The X-SAMPA converter turns the glottalization marker into _>, dunno why

Thooi, chungz ta haix suengz, lamj loonq soonq tiengz noij kuev chungz noz cho hoq nge khoong dwekq tiengz noiz kuev ngweij naaij vyyiz ngweij kie.

[Nortaneous]

As for diacritic-free romanization of a large vowel system, there's Iu Mien:

/i u/ <i u>
/e 3 o/ <e er o>
/{ 6 a: Q/ <ae a aa or>
/ai au a:i a:u Ei @u 3o oi/ <ai au aai aau ei ou eu oi>

[Trebor]

We used to have a Vietnamese member once upon a time, they disappeared though. I can't remember their username. I think there's at least one person who can use it.

Ahh, OK. Do you know of any threads where that learner talks about his/her experience?

Anyway, I guess I'm a little confused as to your goal – because if you actually want to learn Vietnamese and/or communicate in Vietnamese, I'd recommend one of the ASCII options that they generally use to input text, and I'd recommend not trying to change too much, as there might be some passing familiarity with those options among Vietnamese speakers. But if it's just a personal pet project for you to visualize (as it were) the phonology, go ahead.

This task of designing a JAWS-friendly alphabet for the language is just for fun. I don't plan on learning Vietnamese, having too many other things on my to-do list: always more French and Arabic, and tackling Somali, Kinyarwanda, and a few others after.

[Trebor]

Back at this pastime again...

Central Igbo:

Consonants

/p p_h p_j p_j_h t t_h ts\ ts\_h k k_h k_w k_w_h k_<p_</ <p ph py phy t th c ch k kh kw kwh kp/
/b b_t b_j b_j_t d d_t dz\ dz\_t g g_t g_w g_<b_</ <b bh by bhy d dh j jh g gh gw gb>
/f f~ s s~/ /f mf s ns/
/v v~ z z~ G G_w/ <v mv z nz x xw>
/r r~/ <r nr>
/j_0 j_0~ w_0 w_0~ h h~/ <y' ny' w' nw' h nh>
/l j w/ <l y w>

Vowels

/i I e a u U o O/ <ii i e a uu u oo o>

Suprasegmentals

(if mid/downstepped is more common than low)

high tone: acute accent
mid [downstepped?] tone: unmarked
low tone: grave accent

or (if low is more common than mid/downstepped)

high tone: acute accent
mid [downstepped?] tone: circumflex accent
low tone: unmarked

NB: Some of the IPA symbols might have been garbled in the change to CX-SAMPA.

This is one weird phonemic inventory:

- How did the nasalized consonants like /f~/ and /h~/ develop?

- Why do only bilabial, not alveolar or velar, plosives have a four-way distinction?

- How did the voiceless glides arise?

- What happened to /E/?

At least it's a lot easier to invent an orthography for Igbo than for Vietnamese.

[Trebor]

I'm interested in tackling new orthography-creation challenges, so please fire away.

[----]

I'm interested in tackling new orthography-creation challenges, so please fire away.

Here's my conlang Aven:
/p t tS k ?/
/p_h ts_h t_h tS_h k_h/
/p_> ts_> t_> tS_> k_>/
/v D s S s` x h/
/m n n` N/
/m_0 n_0 n`_0/
/l w j r`/
/l_0 w_0 j_0/

/i e { u O @ i~ u~ a~ 7~/
+low, high, and falling/glottalized tone.

[Trebor]

Theta--thanks. Before proceeding, I'd like to ask for the phonotactics.

[----]

sure! All words begin with a consonant or one of the clusters /pj p_hj p_>j kj k_hj k_>j kw k_hw k_>w/. Word medially clusters of up to two consonants are allowed, and only plain stops (or affricates) are allowed before other consonants. Just the same, only plain stops can appear word finally, but there are no limitations on other codas (word final consonants are relatively rare, though). So, possible syllable forms are:
CV
CCV
CVC
CCVC

[Trebor]

Great. That information is very helpful.

Here's my proposal:

Consonants

/p t tS k ?/ <p t c k '>
/p_h ts_h t_h tS_h k_h/ <ph z th ch kh>
/p_> ts_> t_> tS_> k_>/ <p' z' t' c' k'>
/v D s S s` x h/ <v d s š rs x h>
/m n n` N/ <m n rn g>
/m_0 n_0 n`_0/ <m' n' rn'>
/l w j r`/ <l w j r>
/l_0 w_0 j_0/ <l' w' j'>

Vowels

/i e { u O @ i~ u~ a~ 7~/ <i e a u o y iñ uñ añ oñ>
low tone: unmarked
high tone: acute accent (e.g., <í>, <ý>, <óñ>)
falling/glottalized tone: gemination (e.g., <íi>, <ýy>, <óoñ>)

Is there really no /ts/, though?

And how do the voiceless sonorants and falling/glottalized vowels look?

[----]

Is there really no /ts/, though?

And how do the voiceless sonorants and falling/glottalized vowels look?

yeah, plain /ts/ historically merged with /s/, but the aspirated and ejective /ts/ didn't because they behave as fortis sounds.

They look pretty good! Technically writing the voiceless sonorants that way could raise ambiguities but it shouldn't be problematic.

[Nortaneous]

Some conlangs.

Amqoli:
/p_h t_h ts_h tS_h k_h q_h/
/p t ts tS k q~X~R/
/b d dz dZ g/
/s S x h/
/z Z/
/m (n) 5 l_j r r_j/
/a e o i u/

5 l_j and r r_j only contrast before vowels, and r r_j don't contrast before /i/. (The neutralized lateral is written /l/ here.) /m/ is [N] before velars. /n/ is rare. Stress is usually word-initial; I'm only marking non-initial stress here.

bordZu rkaz i ema xadzi lberis. erZel bordZum ndZar_je zgu marideX, erZel bRu taRubza. ema muRrZex, es rbel SqaRori, es bordZum xadzi ne me qol, Zim p_hotsmaSx, Sa obZ"dami regas. Sa bordZu Za temts 5u tXol. Sa xadzi tem, tp_hotsmaSx, topSix, es obZ"dami trega, Sl_ja ham bordZu, 5u, qa"bRux tkel_ji, Sa ndra"p_hotsmaSxam k_hri Nga rugl_jis. rkaz i ema xadzi, ndra"p_hotse seRl_jime erZel temniSri, bordZu tere, Sa 5u taRbri taRbri, Sa 5u buZi. tse mRarZex zgu ga taxenm tSubri.

V'eng:
/p t S~tS k ?/
/b d z~dZ/
/f s S x h/
/m n N/
/B r j G gL\/
/a E O I U i u/

jI? E s jah GU ru s vIN mka. ?a thEN gU ?oS gL\EG, ?a xI s rE E sinar, ?a sitsir Ga. thEN BEN dZi, hI tSa ?Ba riG, hI tSa p?juZ. mjan riG UN hna, mjan dur UN ntI. ?a BEN, hI tSa ?BO dZi brU, hI tSa brU E s tI ?BO tOG a mE gL\EgL\E dZEN rI?. ?O kOm p?ra gL\EG, dZi brU, dZi tOG sI ndI thEN Bo.

[Trebor]

Theta--thanks for the clarification about /ts/. If writing the voiceless sonorants as consonant+apostrophe might cause ambiguity, I would transcribe them double instead--e.g., /w_0/ <ww>.

I didn't really want to impose a central/eastern European feel on your conlang by making /j/ <j>. But it seemed best to avoid an oddity like /j_0@_F_?/ <y'ýy>!

[Trebor]

Some conlangs.

Great.

Amqoli

Consonants

/p_h t_h ts_h tS_h k_h q_h/ <ph th tsh ch kh qh>
/p t ts tS k q~X~R/ <p t ts c k q>
/b d dz dZ g/ <b d dz j g>
/s S x h/ <s sh x h>
/z Z/ <z zh>
/m (n) 5 l_j r r_j/ <m (n) l ly r ry>

Vowels

/a e o i u/ <a e o i u>

Stress

initial: unmarked
non-initial: acute accent

/bordZu rkaz i ema xadzi lberis. erZel bordZum ndZar_je zgu marideX, erZel bRu taRubza. ema muRrZex, es rbel SqaRori, es bordZum xadzi ne me qol, Zim p_hotsmaSx, Sa obZ"dami regas. Sa bordZu Za temts 5u tXol. Sa xadzi tem, tp_hotsmaSx, topSix, es obZ"dami trega, Sl_ja ham bordZu, 5u, qa"bRux tkel_ji, Sa ndra"p_hotsmaSxam k_hri Nga rugl_jis. rkaz i ema xadzi, ndra"p_hotse seRl_jime erZel temniSri, bordZu tere, Sa 5u taRbri taRbri, Sa 5u buZi. tse mRarZex zgu ga taxenm tSubri./

<borju rkaz i ema xadzi lberis. erzhel borjum njarye zgu marideq, erzhel bqu taqubza. ema muqrzhex, es rbel shqaqori, es borjum xadzi ne me qol, zhim photsmashx, sha obzhdámi regas. sha borju zha temts lu tqol. sha xadzi tem, tphotsmashx, topshix, es obzhdámi trega, Slya ham borju, lu, qabqúx tkelyi, sha ndraphótsmashxam khri nga ruglyis. rkaz i ema xadzi, ndraphótse seqlyime erzhel temnishri, borju tere, sha lu taqbri taqbri, sha lu buzhi. tse mqarzhex zgu ga taxenm cubri.>

V'eng

Consonants

/p t S~tS k ?/ <p t x k '>
/b d z~dZ/ <b d j>
/f s S x h/ <f s sh kh h>
/m n N/ <m n g>
/B r j G gL\/ <v r y gh gl>

Vowels

/a E O I U i u/ <a e o i u í ú>

/jI? E s jah GU ru s vIN mka. ?a thEN gU ?oS gL\EG, ?a xI s rE E sinar, ?a sitsir Ga. thEN BEN dZi, hI tSa ?Ba riG, hI tSa p?juZ. mjan riG UN hna, mjan dur UN ntI. ?a BEN, hI tSa ?BO dZi brU, hI tSa brU E s tI ?BO tOG a mE gL\EgL\E dZEN rI?. ?O kOm p?ra gL\EG, dZi brU, dZi tOG sI ndI thEN Bo./

(Why is there a /v/ here?)

<yi' e s yah ghu rú s vig mka. 'a theg gu 'osh glegh, 'a khi s re e sínar, 'a sítsír gha. theg veg jí, hi ca 'va rígh, hi ca p'yúj. myan rígh ug hna, myan dúr ug nti. 'a veg, hi ca 'vo jí bru, hi ca bru e s ti 'vo togh a me glegle jeg ri'. 'o kom p'ra glegh, jí bru, jí togh si ndi theg vo.>

These both look particularly unusual. Did any natlangs inspire Amqoli and V'eng?

[Nortaneous]

That Amqoli romanization is almost identical to the one I use -- the only differences are /ts_h tS_h tS/ <tsx chx ch>. (<c> isn't used, but it used to be /ts/.) It also doesn't mark stress, and always writes <l> before <i> since the 5/l_j contrast there only exists across morpheme boundaries. (So bgulim man-GEN.PL is [bgu5im], but a root like geli is [gel_ji].)

Amqoli was based on Georgian, but it's become its own thing. (It's closely related to another language, Deghuri, that's phonologically based on Hebrew.) V'eng (any /v/ that's there should be /B/) is Tibetan in Vanuatu: seven-vowel system, velar lateral affricate, and preinitials.

Another conlang: Kannow

/p_h t_h t_w_h ts_h ts_w_h tK_h ts`_h c_h k_h k_w_h q_h q_w_h/
/p_> t_> t_w_> ts_> ts_w_> tK_> ts`_> c_> k_> k_w_> q_> q_w_>/
/b d d_w dK\ dz` J\ g g_w G\ G\_w/
/m n n_w (J) N_w r r_w (j) w/
/h/
/a e @ i 1 u/
/a: e: @: i: 1: u:/

Aspirated and voiced plosives may be realized as voiceless and voiced fricatives.
Phonemes in parentheses only appear in loanwords.
Vowel harmony based on height: a e @ ~ i 1 u. Blocked by consonant clusters.
If one consonant in a cluster is rounded, the whole cluster is rounded.
/1/ can't appear after a rounded cluster, and /u/ can't appear after an unrounded cluster.

q_he:remt_>and it_hk_>1nit_hu itK_hip_hnu, hetK_ht_w_hik_hdind, muts`_hq_hdut_w_>aw k_w_hiJ\indiJ\a. t_w_hp_he:nats_ht_>and k_handar. hedK\et_ht_>end, p_htK_>uq_hdund1nt_>adz` k_w_hiJ\indiJ\a rak_we:ts_ht_h ts_h1c_hits_>is_w. hedaq_>at_hts`_h, k_hneN_wew. Nen_wdz`ew, get_hk_>@mg, Nec_htK_>amk_>. hek_w_hardand, Nec_htK_>ammat_> mag_wep_>. hetK_ht_w_hik_hk_hadz`, g_wa:t_>a:c_hk_hadz`, m1k_h1p_hg_wind@nts_hand ts_h1c_hits_>its_w_h i G\a:ts`_>amdz`.

[Trebor]

That Amqoli romanization is almost identical to the one I use -- the only differences are /ts_h tS_h tS/ <tsx chx ch>. (<c> isn't used, but it used to be /ts/.)

Great minds think alike. :p

It also doesn't mark stress, and always writes <l> before <i> since the 5/l_j contrast there only exists across morpheme boundaries. (So bgulim man-GEN.PL is [bgu5im], but a root like geli is [gel_ji].)

I suppose not all conlang orthographies have to be unambiguous. Your decision about [l]-like sounds makes sense though.

Amqoli was based on Georgian, but it's become its own thing. (It's closely related to another language, Deghuri, that's phonologically based on Hebrew.) V'eng (any /v/ that's there should be /B/) is Tibetan in Vanuatu: seven-vowel system, velar lateral affricate, and preinitials.

Cool. No wonder they look so odd.

Now, on to Kannow:

Consonants

/p_h t_h t_w_h ts_h ts_w_h tK_h ts`_h c_h k_h k_w_h q_h q_w_h/ <ph th twh tsh tshw tlh x c kh khw qh qhw>
/p_> t_> t_w_> ts_> ts_w_> tK_> ts`_> c_> k_> k_w_> q_> q_w_>/ <p' t' tw' ts' tsw' tl' x' c' k' kw' q' qw'/
/b d d_w dK\ dz` J\ g g_w G\ G\_w/ <b d dw dl z j g gw qg qgw>
/m n n_w (J) N_w r r_w (j) w/ <m n nw (ny) ngw r rw (y) w>
/h/ <h>

Vowels

/a e @ i 1 u/ <a e ë i ï u>
/a: e: @: i: 1: u:/ <aa ee ëë ii ïï uu>

If only /@/ or /1/ were /o/... then I could avoid those double vowels, which I suspect may not be very pleasant to look at. Then again, <tlh> and <qg> may not be, either.

/q_he:remt_>and it_hk_>1nit_hu itK_hip_hnu, hetK_ht_w_hik_hdind, muts`_hq_hdut_w_>aw k_w_hiJ\indiJ\a. t_w_hp_he:nats_ht_>and k_handar. hedK\et_ht_>end, p_htK_>uq_hdund1nt_>adz` k_w_hiJ\indiJ\a rak_we:ts_ht_h ts_h1c_hits_>is_w. hedaq_>at_hts`_h, k_hneN_wew. Nen_wdz`ew, get_hk_>@mg, Nec_htK_>amk_>. hek_w_hardand, Nec_htK_>ammat_> mag_wep_>. hetK_ht_w_hik_hk_hadz`, g_wa:t_>a:c_hk_hadz`,
m1k_h1p_hg_wind@nts_hand ts_h1c_hits_>its_w_h i G\a:ts`_>amdz`./

<qheeremt'and ithk'ïnithu itlhiphnu, hetlhthwikhdind, muxhqhdutw'aw khwijindija. thwpheenatsht'and khandar. hedletht'end, phtl'uqhdundïnt'az khwijindija rakweetshth tshïcits'isw. hedaq'athxh, khnengwew. ngenwzew, gethk'ëmg, ngectl'amk'. hekhwardand, ngectl'ammat' magwep'. hetlhthwikhkhaz, gwaat'aackhaz, mïkhïphgwindëntshand tshïcits'itshw i qgaax'amz.>

This is sure one language where becoming even half-decent at pronunciation alone would be a tall order for most second-language learners. Was it inspired by something Caucasian or North American? How did you handle all those consonants, and the odd vowels?

[Nortaneous]

It was originally a protolang for a language with a much simpler inventory. (/p t ts tS k f T s S x m n l r v/ + whistled sibilants, /a e @ i u/ + long vowels and diphthongs) I forget what the reasoning was behind Kannow's massive array of POAs, but originally it had a vertical vowel system and /p t t` k/ + palatalization and velarization. Once I added ejectives, I realized that it didn't need to have phonemic fricatives.

The orthography I use is:
Ejective: <pʼ tʼ sʼ lʼ rʼ cʼ kʼ qʼ>
Aspirate: <p t s s ł ř c k q>
Voiced: <b d l r j g ģ>
/m n N/ <m n ņ>
/r w h/ <ŗ w h>

(/tK_h/ is barred l, /ts`_h/ is r-caron, /G\ N r/ are g-, n-, and r-cedilla, and ejectives are written with a following curly quote)

Rounding is written with a grave on the following vowel. Word-final rounded consonants take an acute on the preceding vowel, and an acute plus a grave is a circumflex. Vowels are <a e o i y u>.

There's a lot of allophony, so the first sentence would be pronounced something like: [X{:remt_>And ITk_>1n_jiTu itK_hifnu heKt_hyxdInd mus`q_hdut_>Ow koi_^j\indIi_^j\a] And the ejectives are weak -- the salient feature isn't ejectivity, but that they're always realized as unaspirated, unvoiced stops. They turn to lenis stops in the descendant, whereas aspirates turn to fricatives or affricates and voiced stops lenite to resonants.

(I don't have anything on the descendant yet, since the sound changes give it more irregularity in its verbal system than Old Irish, so massive analogy is guaranteed.)