zompist bboard

So I've got a phonology here that I've given a Romanization to. I feel like I've done the best I could, but I see some shortcomings of it, and I want to know if it's good or whether it could use some further tweaking.

Right now, I've got a system that draws inspiration from Czech, Sami, and (for the vowels) Vietnamese with some slight reassignments. In what's supposed to be the alphabetical order for this language:
<a è e ö i j ü ơ ư y ò o u w p f b v m t ŧ ť d đ ď s š z ž n k g h ǥ ŋ l r>
/a ɛ e ø i j y ɤ ɯ ɰ ɔ o u w p f b v m t θ t͡ʃ d ð d͡ʒ s ʃ z ʒ n k ɡ x ɣ ŋ l r/

(If you're wondering where /ɥ/ is, it's actually [ɥ], an allophone of /j/ before rounded vowels. )

I've got a (C)(j/ɰ/w/l/r)V(C) syllable structure with a few phonological restrictions:

• The first C can't be an approximant or liquid. Syllables can start with a liquid, but not an approximant; /j ɰ w/ have become /ʒ ɣ v/ everywhere except after a syllable-initial consonant.
• There's no /tj dj sj zj/, they've become /t͡ʃ d͡ʒ ʃ ʒ/. Likewise, /ti di si zi/ is nearly always /t͡ʃi d͡ʒi ʃi ʒi/; though the former can occur in foreign words, they're usually pronounced as the latter.

There are also supposed to be a couple oddities with native words that is really a product of sound change more than anything else:

• This language is supposed to come from an earlier one where coda consonants couldn't occur after long vowels or diphthongs. Since /ɛ ø y ɔ/ are monophthongizations of /aj oj uj aw/, they're not followed by coda consonants in the native language, though there's usually no trouble with pronouncing coda consonants after /ɛ ø y ɔ/ in foreign words.
• Palatalization is recessive only in the native language, thus /t͡ʃ d͡ʒ ʃ/ don't occur syllable-finally in native words in the standard language (though in some dialects, /t d s z/ /t͡ʃ d͡ʒ ʃ ʒ/ | i_$), and in uneducated speech they often have an /i/ inserted after them when they occur in foreign words (so <öť>, the word I currently have for copper, would be pronounced [ˈø.t͡ʃi] instead of /øt͡ʃ/ in these dialects).

I like how consistent my system is with its diacritics (grave for lowering, umlaut for fronting, hook for unrounding, stroke for fricativization, háček for palatalization), but I'm kinda worried about representing /t͡ʃ d͡ʒ/ as <ť ď>: in Czech, <ť ď> represent /c ɟ/, whereas /t͡ʃ d͡ʒ/ are written <č dž>. My /t͡ʃ d͡ʒ/ don't derive from an earlier /c ɟ/, so I don't have the excuse I do for representing /ɤ ɯ/ as <ơ ư> from earlier /ə ɨ/. Do you think that could pose a problem? If so, what do you recommend, and how should I resolve collisions with other letters, especially the one for /j/?

Here are the alternatives I've considered:

1. /t͡ʃ d͡ʒ j/ are represented as <ť ď j>: the current system, which is consistent and justified, but the first two of which might end up mispronounced by people familiar with Czech.
2. /t͡ʃ d͡ʒ j/ are represented as <č dž j>: more closely follows Czech, but I'd like to have /d.ʒ/ and would need to resolve conflicts with it. That would most likely become <d·ž> in this system.
3. /t͡ʃ d͡ʒ j/ are represented as <č ǰ j>: the alternative I find the most attractive, but I would have to resort to combining diacritics to write <J̌>, so it might cause display issues.
4. /t͡ʃ d͡ʒ j/ are represented as <č j y>: I think the influence here is obvious, but I would need to come up with another letter for /ɰ/. I had <ğ> in earlier versions when there was a more straightforward Turkish influence in my system, but it looks weird to me in the contexts I foresee it occurring despite it making sense to me that it'd be a similar letter to the one for /ɣ/.

What are your opinions about all this?

Phoneme symbols are sort of fake. There are a few languages that are described as contrasting /ɨ ɯ/, but in general the two are used interchangeably. Vietnamese has been described as having /ɨ/ or /ɯ/. It depends on the author.

I don't see why not to use <ty dy y> or <tj dj j>. It adds depth, if nothing else.

/ɰ ɣ/ don't contrast, right? Maybe across morpheme boundaries: *C1C2-ɰ would condition fortition but *C1-C2ɰ wouldn't. I wouldn't bother to distinguish them. Maybe not any of the voiced fricative/approximant pairs.

Nortaneous wrote:/ɰ ɣ/ don't contrast, right? Maybe across morpheme boundaries: *C1C2-ɰ would condition fortition but *C1-C2ɰ wouldn't.

Consonant clusters can't occur in the coda, but yeah. VC-ɰ triggers fortition while V-Cɰ doesn't. I see your point in the approximants really not having much functional load compared to their voiced fricative counterparts, though. It'd certainly simplify the analysis not to consider them different phonemes.

Corrected the OP–/ʒ/ CAN occur syllable-finally natively, just not from /z/ in the standard language.

Could write /ʒ/ as <j> except where it's from palatalization of /z/.

I was thinking of doing that

StrangerCoug wrote: <a è e ö i j ü ơ ư y ò o u w p f b v m t ŧ ť d đ ď s š z ž n k g h ǥ ŋ l r>
/a ɛ e ø i j y ɤ ɯ ɰ ɔ o u w p f b v m t θ t͡ʃ d ð d͡ʒ s ʃ z ʒ n k ɡ x ɣ ŋ l r/

There are six different diacritics here. I would suggest this instead:

<a è e ö i j ü ë ï y ò o u w p f b v m t th tj d dh dj s sj z zj n k g kh gh ŋ l r>
/a ɛ e ø i j y ɤ ɯ ɰ ɔ o u w p f b v m t θ t͡ʃ d ð d͡ʒ s ʃ z ʒ n k ɡ x ɣ ŋ l r/

My rationale for this is:

1. There's no reason to have a separate diacritic for unround back vowels; we can say the diaeresis means "opposite of expected frontness".
2. Given that

   StrangerCoug wrote: There's no /tj dj sj zj/, they've become /t͡ʃ d͡ʒ ʃ ʒ/

   there are good etymological and phonological reasons to write /tʃ dʒ ʃ ʒ/ as <tj dj sj zj>. This also means you can distinguish between /dʒ d.ʒ/ as <dj dzj>.
3. The digraphs are a bit ugly, but <h> provides a consistent way of indicating fricativisation in <th dh kh gh>, and since it appears nowhere else, there's no possibility of confusion.

Buran wrote:

StrangerCoug wrote: <a è e ö i j ü ơ ư y ò o u w p f b v m t ŧ ť d đ ď s š z ž n k g h ǥ ŋ l r>
/a ɛ e ø i j y ɤ ɯ ɰ ɔ o u w p f b v m t θ t͡ʃ d ð d͡ʒ s ʃ z ʒ n k ɡ x ɣ ŋ l r/

There are six different diacritics here. I would suggest this instead:

<a è e ö i j ü ë ï y ò o u w p f b v m t th tj d dh dj s sj z zj n k g kh gh ŋ l r>
/a ɛ e ø i j y ɤ ɯ ɰ ɔ o u w p f b v m t θ t͡ʃ d ð d͡ʒ s ʃ z ʒ n k ɡ x ɣ ŋ l r/

My rationale for this is:

1. There's no reason to have a separate diacritic for unround back vowels; we can say the diaeresis means "opposite of expected frontness".
2. Given that

   StrangerCoug wrote: There's no /tj dj sj zj/, they've become /t͡ʃ d͡ʒ ʃ ʒ/

   there are good etymological and phonological reasons to write /tʃ dʒ ʃ ʒ/ as <tj dj sj zj>. This also means you can distinguish between /dʒ d.ʒ/ as <dj dzj>.
3. The digraphs are a bit ugly, but <h> provides a consistent way of indicating fricativisation in <th dh kh gh>, and since it appears nowhere else, there's no possibility of confusion.

I like yours better than mine I'll adopt it.

Matter of fact, I think I've got a way to combine Buran and Nortaneous's ideas: Make the Romanization system a little more like a transcription system rather than a transliteration system. <v> and <w> are the same letter in the conscript, as are <gh> and <y>. (<tj dj sj zj> I want to make digraphs in the conscript too.) The problem, which I'd really consider minor in practice, is that the conscript wouldn't be 100% phonemic. It distinguishes between affricates and stop-fricative sequences (at the cost of masking the distinction between historical /d.j/ and historical /d.zj/), but doesn't always tell you whether to pronounce <j w/v y/gh> as a fricative or a semi-vowel. I feel like I can live with it in practice, though.

I base my romanisation schemes on the phonology of the original language of the script at the time of the script's invention. Even when the script is borrowed or after the language gives way to descendants I use the same romanisation, which reflects the way alphabets evolved in real life. I could give you the romanisation I would use if you show what the original phonology was like.

OK