Phonology, Transliteration, and Transcription Critique Thread

[StrangerCoug]

I could have sworn we had a thread like this, but it turned out it was the Sound Change Critique Thread I was thinking of and the question I have isn't really a sound change question. I created this thread to have a more appropriate venue for the kind of question I had, and I open this thread to anyone else who also has similar questions about their phonology and how they're Romanizing, Cyrillizing, Arabizing, etc. their sounds.

Let's begin:

I have a conlang with this phonology:

More: show

Consonants

	Labial	Dental	Alveolar	Palatal/Velar	Uvular	Glottal
Nasal	m /m/	nh /n̪/	n /n̺/	ng /ŋ/
Tenuis plosive	b /p/	dh /t̪/	d /t̺/	g /k/	ġ /q/	ʼ /ʔ/
Aspirate plosive	p /pʰ/	th /t̪ʰ/	t /t̺ʰ/	k /kʰ/	q /qʰ/
Ejective plosive	pʼ /pʼ/	thʼ /t̪ʼ/	tʼ /t̺ʼ/	kʼ /kʼ/	qʼ /qʼ/
Tenuis affricate		zh /t̪͡s̪/	z /t̺͡s̺/
Aspirate affricate		ch /t̪͡s̪ʰ/	c /t̺͡s̺ʰ/
Ejective affricate		chʼ /t̪͡s̪ʼ/	cʼ /t̺͡s̺ʼ/
Fricative		sh /s̪/	s /s̺/			h /h/
Approximant/Flap	w /w/	l /l/	r /ɾ/	j /j/

The phonotactics below take care of most of the potential ambiguities in the transliteration of the consonants. The glottal consonants don't like to be postconsonantal, but to force a digraph to be read as a sequence of an alveolar consonant + /h/ instead of a single dental consonant when the possibility of confusion exists, a hyphen is written in between the constituent letters.

Vowels

	Short front	Long front	Short back	Long back
Close	i /i/	ī /iː/	u /u/	ū /uː/
Mid	e /e/	ē /eː/	o /o/	ō /oː/
Open			a /a/	ā /aː/

Six diphthongs also exist: ai /ai̯/, au /au̯/, ei /ei̯/, eu̯ /eu/, oi /oi̯/, and ou /ou̯/.

Phonotactics
C(C)V(C). Syllable-initial clusters are always made up of a non-glottal obstruent followed by a consonant in the approximant/flap row. /j/ cannot immediately precede /i(ː)/, and /w/ cannot immediately precede /u(ː)/ Coda consonants are allowed only after a short monophthong and cannot be glottal, aspirated, or ejective; coda semivowels do not contrast with the falling element of the diphthongs and are transcribed as full vowels. Coda occlusives assimilate to following obstruents except that /p/, /k/, and /q/ do not assimilate to a following sibilant. Any allowed coda consonant can be geminated.

Except for the liquids, consecutive coronals must be either all dental or all alveolar; even then, /ɾ/ cannot immediately follow a dental consonant.

The phonotactics are planned to be a bit more complex than this in morpheme-medial position in native words, and I've hinted at this in the consonants section. I'm still working on the details, but they mostly have to do with euphonics.

I'd like a second opinion about using a following h to transcribe the dental series, an idea I got from the transcription of Australian Aboriginal languages. (The rest is Standard Average Me for languages with tenuis, aspirated, and ejective stops but no phonemic voiced stops; that mostly combines influences from Tlingit and Pinyin, with ġ for /q/ having a loose basis in the ISO romanization of Arabic.) If the alveolar consonants were more common than the corresponding dental consonants, I'd be fine with it, but I want it to be true only for the nasal and sibilant, and I'd rather not write the more common of two similar sounds as a digraph and the less common of them as a monograph.* I thought of using diacritics for the alveolar series, but then the hangup becomes using a diacritic for the more common sibilant. I know Basque writes /s̪/ as z and /s̺/ as s, but here z is taken for /t̺͡s̺/.

Thoughts? Ideas for improvement?

* That said, my transcription system (as opposed to my transliteration system) is fine with omitting the h where the phonotactics leave no room for confusion; a similar simplification it allows is writing ng as just n when there's no room for confusion there, either.

[StrangerCoug]

I've just decided that sequences of a coronal stop and a sibilant don't contrast with geminate affricates, so I actually might be able to get away with this at the expense of the transliteration of /t̺͡s̺/ looking slightly odd (the other cells are the same as before for the time being):

	Dental	Alveolar
Tenuis affricate	dz /t̪͡s̪/	ds /t̺͡s̺/
Aspirated affricate	tz /t̪͡s̪ʰ/	ts /t̺͡s̺ʰ/
Ejective affricate	tzʼ /t̪͡s̪ʼ/	tsʼ /t̺͡s̺ʼ/
Fricative	z /s̪/	s /s̺/

What do you think?

Edit: OK, that should be right now.

[Moose-tache]

Adding h in Mandarin does kind of the opposite, turning alveolars into retroflex. But only for affricates, which you seem to have solved already. Doing it for plosives strikes me as a little weird. You could use C and J for dentals, but... nah, that's no good either. I think the only way to do this in a way that doesn't look counter-intuitive is to add another diacritic. Unfortunately.

[StrangerCoug]

I don't like having c j represent dentals, either, even if I reassign /j/ to y to free up j.

I don't think that leaving the h for the remaining dentals is the worst solution in hindsight (some dialects of English pronounce the dental fricatives as dental stops, if I want a justification), but I agree that I will likely have to resort to diacritics if my goal is to write the more common letters with simpler-to-type forms. Which in turn likely means another compromise in which coronal /n/ is more likely to be written with a diacritic than without it, though it has a partial counterbalance.

Perhaps this, taking a cue from the transcription of Malayalam?

	Dental	Alveolar
Nasal	n /n̪/	ṉ /n̺/
Tenuis stop	d /t̪/	ḏ /t̺/
Tenuis affricate	t /t̪ʰ/	ṯ /t̺ʰ/
Tenuis ejective	tʼ /t̪ʼ/	ṯʼ /t̺ʼ/
Tenuis affricate	dz /t̪͡s̪/	ds /t̺͡s̺/
Aspirated affricate	tz /t̪͡s̪ʰ/	ts /t̺͡s̺ʰ/
Ejective affricate	tzʼ /t̪͡s̪ʼ/	tsʼ /t̺͡s̺ʼ/
Fricative	z /s̪/	s /s̺/

If I do this, I'll probably have /n̺.n̺/ consistently written ṉṉ for consistency with the way the other letters written as monographs are geminated, but in transcription I can otherwise allow the underbar to be dropped when not needed (e.g. what in transliteration would have to be ṉḏr can be just ndr since I don't allow r after dentals). It's also consistent with not adding underbars under the alveolar affricates.

Edit: Corrected the IPA transcriptions for the plosives to match the row and column headers.

[Travis B.]

Adding h in Mandarin does kind of the opposite, turning alveolars into retroflex. But only for affricates, which you seem to have solved already. Doing it for plosives strikes me as a little weird. You could use C and J for dentals, but... nah, that's no good either. I think the only way to do this in a way that doesn't look counter-intuitive is to add another diacritic. Unfortunately.

If anything, <c> and <j> seem more appropriate for alveolars than dentals.

[StrangerCoug]

Adding h in Mandarin does kind of the opposite, turning alveolars into retroflex. But only for affricates, which you seem to have solved already. Doing it for plosives strikes me as a little weird. You could use C and J for dentals, but... nah, that's no good either. I think the only way to do this in a way that doesn't look counter-intuitive is to add another diacritic. Unfortunately.

If anything, <c> and <j> seem more appropriate for alveolars than dentals.

There's also precedent in Kiowa for using j as a coronal stop (I checked a few moments ago), but it uses c as a velar stop. It also doesn't change the fact that using c j for stops pronounced that far front seems unusual to me, even if it is in between what I'm using for t d and what I'm using for k g.

[chris_notts]

As you say, h digraphs to mark dental place are pretty standard for Australian languages, but I have to admit that I don't find it very aesthetically pleasing, and the j digraphs for palatals also don't look that nice. I think in the case of Australian languages, part of the dislike comes from the fact that many of these languages don't have massive phonologies, it's just that the Roman alphabet with something like its usual symbol to sound mapping isn't very suited to writing them.

[StrangerCoug]

OK, I've decided I don't like how the underbar looks on consecutive consonants (I have to use such a sequence in the conlang's name), so let's try an underdot instead and see how that looks.

[StrangerCoug]

How's this revision?

Consonants

	Labial	Dental	Alveolar	Palatal/Velar	Uvular	Glottal
Nasal	m /m/	n /n̪/	ṇ /n̺/	ng /ŋ/
Tenuis plosive	b /p/	d /t̪/	ḍ /t̺/	g /k/	ġ /q/	ʼ /ʔ/
Aspirate plosive	p /pʰ/	t /t̪ʰ/	ṭ /t̺ʰ/	k /kʰ/	q /qʰ/
Ejective plosive	pʼ /pʼ/	tʼ /t̪ʼ/	ṭʼ /t̺ʼ/	kʼ /kʼ/	qʼ /qʼ/
Tenuis affricate		dz /t̪͡s̪/	ds /t̺͡s̺/
Aspirate affricate		tz /t̪͡s̪ʰ/	ts /t̺͡s̺ʰ/
Ejective affricate		tzʼ /t̪͡s̪ʼ/	tsʼ /t̺͡s̺ʼ/
Fricative		z /s̪/	s /s̺/			h /h/
Approximant/Flap	w /w/	l /l/	r /ɾ/	j /j/

Vowels

	Short front	Long front	Short back	Long back
Close	i /i/	ī /iː/	u /u/	ū /uː/
Mid	e /e/	ē /eː/	o /o/	ō /oː/
Open			a /a/	ā /aː/

[WeepingElf]

That's fine to me.

[bradrn]

Fine with me too. I don’t love the transcription of the affricates/fricatives, but really there’s few good options there.

[chris_notts]

Fine with me too. I don’t love the transcription of the affricates/fricatives, but really there’s few good options there.

Basque uses z/s/x for unvoiced sibilant fricative and affricate place distinctions and I actually quite like it.

[bradrn]

Fine with me too. I don’t love the transcription of the affricates/fricatives, but really there’s few good options there.

Basque uses z/s/x for unvoiced sibilant fricative and affricate place distinctions and I actually quite like it.

Yes, I know; I really dislike it. (I was going to mention Basque, but I saw someone already did.)

[chris_notts]

Fine with me too. I don’t love the transcription of the affricates/fricatives, but really there’s few good options there.

Basque uses z/s/x for unvoiced sibilant fricative and affricate place distinctions and I actually quite like it.

Yes, I know; I really dislike it. (I was going to mention Basque, but I saw someone already did.)

Another similar thing I've copied from reference books about Mayan, but I don't fully understand the motivation for, is the tendency to use both s and z for the same place though. Specifically, to write the fricative s but the corresponding affricate tz.

[Ares Land]

Basque uses z/s/x for unvoiced sibilant fricative and affricate place distinctions and I actually quite like it.

Yes, I know; I really dislike it. (I was going to mention Basque, but I saw someone already did.)

Another similar thing I've copied from reference books about Mayan, but I don't fully understand the motivation for, is the tendency to use both s and z for the same place though. Specifically, to write the fricative s but the corresponding affricate tz.

Maybe an attempt to rationalize spelling conventions borrowed from Spanish?
The usual convention for Nahuatl is also to use <tz> for the affricate, but the fricative is z or c, or sometimes ç. <c> is of course also used for /k/, which is written qu before a front vowel, and for the labio-velar as <cu> or <uc>

[chris_notts]

Yes, I know; I really dislike it. (I was going to mention Basque, but I saw someone already did.)

Another similar thing I've copied from reference books about Mayan, but I don't fully understand the motivation for, is the tendency to use both s and z for the same place though. Specifically, to write the fricative s but the corresponding affricate tz.

Maybe an attempt to rationalize spelling conventions borrowed from Spanish?
The usual convention for Nahuatl is also to use <tz> for the affricate, but the fricative is z or c, or sometimes ç. <c> is of course also used for /k/, which is written qu before a front vowel, and for the labio-velar as <cu> or <uc>

I was wondering if it was a handwriting thing? Tz is nicer to write cursively because you can go straight from the cross on the t to the top of the z without a sharp direction reversal. With s you shoot over then reverse course sharply to curve back again.

[Oxygenman]

I'm designing a Dwarven conlang for a friend's long-standing D&D campaign world, based off of a handful of place and personal names he has used in his campaigns over the years. I want to be as faithful as possible to his "romanization" system (i.e. what looked "Dwarven" to a bunch of 13-year-olds over two decades ago), and am looking for ideas how to reconcile some phonology/romanization roadblocks. As noted below, the best I've been able to do it say that some words just retain "archaic" spellings but I'd like to know of any thoughts for "cleaner" solutions.


1. h's follow voiced and voiceless plosives, e.g. "Bharn" and "Tharn"
To remain faithful to our own pronunciation of such words, I've used th and dh, to represent /θ/ and /ð/, respectively, and by extension decided kh and gh would be /x/ and /ɣ/, respectively. He definitely used ph to represent /f/, but we always pronounced bh as just /b/. I had these sounds evolve from a series of "voiced aspirates" but I'm not sure how to justify bh sticking around in the romanization system. The only thing I've been able to think up is a pseudo in-game explanation that that the romanization system contains some archaic spellings


2. h's precede sonorants, e.g. "Bahn "Mahr"
I know "pre-aspiration" is potentially a thing here, but I understand that it is diachronically unstable. Given that this language also has long vowels (see below), I'm not convinced it would be able to maintain phonetic pre-aspiration and vowel length. Again, my only thought it is say these are basically just long vowels but were once pre-aspirated and use archaic spelling.


3. Circumflexes over vowels, e.g. Gôr, Dân
I decided this would be phonetic vowel length.


4. Uses both K and C, e.g. Can, Rukur
To remain faithful to our own pronunciation of such words, again, my only thought it is say c was once a palatal /c/ that merged with the velar series so now both are /k/ but some words still retain archaic spelling.

[bradrn]

1. h's follow voiced and voiceless plosives, e.g. "Bharn" and "Tharn"
To remain faithful to our own pronunciation of such words, I've used th and dh, to represent /θ/ and /ð/, respectively, and by extension decided kh and gh would be /x/ and /ɣ/, respectively. He definitely used ph to represent /f/, but we always pronounced bh as just /b/. I had these sounds evolve from a series of "voiced aspirates" but I'm not sure how to justify bh sticking around in the romanization system. The only thing I've been able to think up is a pseudo in-game explanation that that the romanization system contains some archaic spellings

Alternately, you could say that ⟨Ch⟩ is the orthographic representation of a consonant mutation system (which is a convention I’ve used before). In this case, perhaps ⟨bh⟩ was previously a /β/ which got fortited back to /b/, but the consonant mutation is still represented even in cases where it’s phonologically invisible. It would be a bit weird to have /ɸ/→/f/ but /β/→/b/, but it at least avoids archaisms in the romanisation.

2. h's precede sonorants, e.g. "Bahn "Mahr"
I know "pre-aspiration" is potentially a thing here, but I understand that it is diachronically unstable. Given that this language also has long vowels (see below), I'm not convinced it would be able to maintain phonetic pre-aspiration and vowel length. Again, my only thought it is say these are basically just long vowels but were once pre-aspirated and use archaic spelling.

These could also denote voiceless sonorants: /ban̥ mar̥/.

4. Uses both K and C, e.g. Can, Rukur
To remain faithful to our own pronunciation of such words, again, my only thought it is say c was once a palatal /c/ that merged with the velar series so now both are /k/ but some words still retain archaic spelling.

You could also take zompist’s convention for Verdurian: ⟨c k⟩ is /k q/. Minimally different from what you’re doing currently, yet justified. (Though, for that matter, /c k/ would itself be minimally different from what you’re doing currently.)

An alternate cop-out would be to say that the two sounds have merged in this dialect, but remain separate in others so are written differently.

[Lērisama]

1. h's follow voiced and voiceless plosives, e.g. "Bharn" and "Tharn"

Maybe it represented glottalisation of some kind? It would make sense for the voiced plosives – if voiced plosive+h were implosives then it would make sense for only the back plosives to lenite, as implosion is more stable on front plosives, and then the solitary *ɓ would be more vulnerable to lose its implosion. According to the index diachronia, in Black Thai earlier *ɓ *ɗ → /b l/¹, so I'd believe *ɓ *ɗ *ɠ → /b ð ɣ/. Annoyingly glottalisation on voiced plosives tends to correspond to a lack of aspiration on voiceless plosives, but I'd believe a route of preglottalised→geminated→aspirated→fricatives. This would work well with Bradrn's consonant mutation idea: pre-dwarven *-ʔ barn and *-ʔ tarn or some such could become *ɓarn *tːarn → modern /barn θarn/

2. h's precede sonorants, e.g. "Bahn "Mahr"

Maybe Bradrn's idea of voiceless sonorants, or furthering the ⟨h⟩ = glottalisation association I suggested earlier, maybe creaky voice i.e. /ban̰ mar̰/

4. Uses both K and C, e.g. Can, Rukur

I don't know if it'd work with all the words you already have, but maybe do an (old) Latin, and have the choice of ⟨c⟩ and ⟨k⟩ based on the preceding vowel².

¹ It doesn't seem to have had *ɠ
² Although the other way round: Old Latin would have had ⟨KAN⟩ and ⟨RUCUR⟩

[Nortaneous]

1. h's follow voiced and voiceless plosives, e.g. "Bharn" and "Tharn"
To remain faithful to our own pronunciation of such words, I've used th and dh, to represent /θ/ and /ð/, respectively, and by extension decided kh and gh would be /x/ and /ɣ/, respectively. He definitely used ph to represent /f/, but we always pronounced bh as just /b/. I had these sounds evolve from a series of "voiced aspirates" but I'm not sure how to justify bh sticking around in the romanization system. The only thing I've been able to think up is a pseudo in-game explanation that that the romanization system contains some archaic spellings

Written Dwarvish preserves a lost contrast between /b/ and an implosive /ɓ/.

2. h's precede sonorants, e.g. "Bahn "Mahr"
I know "pre-aspiration" is potentially a thing here, but I understand that it is diachronically unstable. Given that this language also has long vowels (see below), I'm not convinced it would be able to maintain phonetic pre-aspiration and vowel length. Again, my only thought it is say these are basically just long vowels but were once pre-aspirated and use archaic spelling.

Marker of now-lost tone derived from *-Rs clusters? A tonal contrast in long vowels?

4. Uses both K and C, e.g. Can, Rukur
To remain faithful to our own pronunciation of such words, again, my only thought it is say c was once a palatal /c/ that merged with the velar series so now both are /k/ but some words still retain archaic spelling.

Written Dwarvish preserves a lost contrast between /k/ and an implosive *ɠ > ƙ. *ɗ had already been lost by the time of the earliest known texts.