Luworese rounding harmony and other morphophonology

Substantial postings about constructed languages and constructed worlds in general. Good place to mention your own or evaluate someone else's. Put quick questions in C&C Quickies instead.
Post Reply
User avatar
Chengjiang
Avisaru
Avisaru
Posts: 437
Joined: Sun Feb 26, 2006 4:41 am
Location: Davis, CA

Luworese rounding harmony and other morphophonology

Post by Chengjiang »

Right now I'm remaking my old developmentally arrested conlangs from a few years back. One of these is Luworese. I'd like some input regarding its rather extensive morphophonological processes. One of my design goals with this language was always to make a relatively low-fusion language that looks more fusional than it is due to the wide range of possible allomorphs of many morphemes. (Luworese is a very high-synthesis, possibly polysynthetic language; I'm never quite sure where the dividing line is.) One of my other design goals was just to create a language with a unique sound that was plausible but didn't strongly resemble any one natural language.

I'll discuss the main morphophonological alternations over a few posts. This post will simply describe the base phonology.

Consonant inventory

Luworese is characterized by distinguishing a large number of consonants without having more than a marginal voicing contrast, or anything analogous such as aspiration or glottalization. The consonants are as follows, followed by their usual orthographic representation.

Nasal: /m n ɲ ŋ/ m n ny ng
Stop: /p t c k kʷ ʔ/ p t ky k kw '/t
Affricate: /ts tɬ tʃ (cʎ̥)/ ts tl ch (tly)
Fricative: /ɸ s ɬ ʃ (ʎ̥) h/ f s hl sh (hly) h
Approximant: /B D l j (ʎ) ɰ w/ b d l y (ly) g w

Among the palatals, the stop is a solidly dorsal "true" palatal while the nasal and laterals vary between dorsal and alveopalatal. The palatal laterals are normally analyzable as underlyingly the alveolar laterals in palatalizing environments, with there being a single morpheme that seems to contain an inherently palatal lateral consonant. Also, the palatal lateral obstruents do stay fricative: I just didn't feel like putting two diacritics on a character that's going to keep showing up.

The glottal stop is represented as t when it occurs word-finally for reasons that will be discussed with further information on morphophonology. /c/ is written k before /i/, as other dorsal stops do not occur in that environment. /kʷ/ is written k before /u/ or /o/ for similar reasons, although it can still contrast with /c/ there. Any geminated consonant whose normal orthographic representation has two letters doubles the first letter only, e.g. /ʃː/ is written ssh.

The consonants represented /B D/ here are actually something that I'm still uncertain about. Their properties are similar to Finnish /d/: They originated as weakened versions of /p t/ (with the various dorsal glides being the weakened forms of the dorsal stops), and unlike the other approximants they cannot occur word-initially. I've given them vague representations here because I'm not actually sure how I want to realize them in this new version of the language. I tentatively have a (compressed, not rounded) bilabial approximant for /B/ and an alveolar tap (not treated as a rhotic) for /D/, but I'm not 100% sure that's what I want to go with. They can only occur intervocalically.

Vowel inventory

Monomoraic: /i ɯ u ɛ ʌ ɔ a/ i ư u e ơ o a
Dimoraic: /iː ɯː uː ɛː ɛi ɛɯ ɛu ʌː ʌi ʌɯ ɔː ɔi ɔu aː ai aɯ au/ ii ưư uu ee ei eư eu ơơ ơi ơư oo oi ou ai aư au

All diphthongs are falling.

Stress and timing

Luworese has moraic timing, with short syllables (short vowel, no coda) taking up one mora and long syllables (long vowel or short plus coda) taking up two. Luworese has pitch-based stress characterized by a downstep that I think will be marginally contrastive, although I haven't worked out the details yet.

Phonotactics

Luworese allows syllables of the form (C)V(V/F), where V equals one mora of vowel, VV must be one of the above legal dimorae, and F is one of a restricted set of finals. Word-internal vowel-vowel hiatus typically gets eliminated in the surface representation by glide insertion or diphthong formation, but the details of that are complex and will be addressed later.

Legal finals, sticking relatively close to the surface representation and not getting too deep into archiphonemes tied to morphophonological processes: /m N ɲ ŋ p T k s h l/

/N/ is a nasal that assimilates to the POA of a following consonant and is realized as [n] before a glottal consonant or a pause. It is normally written n. Since /m ɲ ŋ/ are also possible finals, what this means is that /m ɲ ŋ/ can precede a heterorganic following consonant but /n/ cannot. /T/ is a stop that is similar to /N/; it assimilates to the POA of a following consonant and is realized as [ʔ] before a pause. It has some complex interactions with fricatives and approximants as far as what's actually produced, but for now let's just say that while all obstruents except /ʔ/ can be geminated, /t/ can't precede a heterorganic consonant while /p k/ can.

Major allophonic rules not covered above

Coda stops lack an audible release, and if they aren't part of a geminate they tend to be lightly glottalized. Coda /m p/ often have a velar component and may approach coarticulated /ŋm kp/ if they aren't part of a geminate. /l/ is flapped [ɺ] intervocalically, and tends to centralize to [ɹ] before central coronals or before a pause.

There's a hell of a lot more about how consonant clusters end up resolving, but that's for another post.

Next up, we'll be covering rounding harmony, a process that affects much more than just the vowels in this language and, as an echo of the diachronics behind the present phonology, affects some consonants that are neither rounded nor have rounded counterparts.
[ʈʂʰɤŋtɕjɑŋ], or whatever you can comfortably pronounce that's close to that

Formerly known as Primordial Soup

Supporter of use of [ȶ ȡ ȵ ȴ] in transcription

It is pitch black. You are likely to be eaten by a 青.

User avatar
Chengjiang
Avisaru
Avisaru
Posts: 437
Joined: Sun Feb 26, 2006 4:41 am
Location: Davis, CA

Re: Luworese rounding harmony and other morphophonology

Post by Chengjiang »

Rounding harmony: It's not just for vowels anymore

The vowels of Luworese are subject to rounding harmony. With very few exceptions, any morpheme's vowels, and any suffixes found after a given lexical morpheme, must draw from one of these two sets:

Unrounded: /i ɯ ɛ ʌ a/
Rounded: /i u ɛ ɔ a/

These include the vowels making up dimorae, e.g. /eɯ/ can only occur in unrounded harmonic environments. As you can see, /i ɛ a/ are neutral, with the harmony applying to the back vowels.

Velar consonants

Velar consonants are also divided into unrounded and rounded sets that are subject to harmony:

Unrounded: /ŋ k ɰ/
Rounded: /ŋ kʷ w/

/ŋ/ is neutral (in onsets; for codas see below), while oral velar consonants must conform to the rounding of their environment. Thus morphemes like -kweno- and -gaka- are possible but *-kwenơ- and *-waka- are not. Keep in mind that sequences like -kuhl- actually have /kʷ/, not /k/. Palatal consonants and all non-dorsals are neutral.

Coda nasals

Nasal codas in Luworese are divided into "strong" and "weak". Weak nasal codas contain underlying /N/: They surface with a nasal homorganic to the following consonant, or [n] before a glottal or a pause. Strong nasal codas contain one of underlying /m ɲ ŋ/, and surface with that nasal.

Weak nasal codas are neutral to rounding harmony and can occur in all environments. Strong nasal codas are restricted to only occurring after certain vowels:

Palatal: /iɲ ɛɲ ɔɲ ʌɲ aɲ/
Velar: /ɯŋ ɛŋ ʌŋ aŋ/
Labial: /um ɛm ɔm am/

Note that on the surface, coda nasals can still come after the "wrong" vowel if they are homorganic to the next syllable's onset. -tsempư- is possible; *-tsemtư- is not. Keep in mind also that this only affects the coda: Onset /m/ can occur in unrounded environments and onset /ŋ/ can occur in rounded environments.

Strong nasal codas behave like the second element of diphthongs for the purposes of harmony. The velar set can only occur in unrounded environments and the labial set can only occur in rounded environments. This means that not only is, for example, /um/ restricted to rounded environments; /ɛm/ and /am/ are as well. While the palatal set is neutral, I've included it here so it's clear which vowels can be followed by strong /ɲ/. (Anything that's not a close back vowel.)

Coda stops

Like coda nasals, coda stops can be divided into strong and weak varieties. The weak coda stop is /T/, which surfaces as a homorganic stop to the following consonant (subject to some alterations like affrication of fricatives which will be discussed in more detail later) or as [ʔ] before a pause. The strong coda stops are /p k/, which can occur before consonants of different POA or before a pause.

As with /N/, /T/ can occur in all environments. Strong /p k/ pattern like /m ŋ/:

Velar: /ɯk ɛk ʌk ak/
Labial: /up ɛp ɔp ap/

The velar set only occurs in unrounded environments, the labial only in rounded. As before, coda [p] and [k] can still occur outside of the expected environment if they precede a consonant of the same POA, because they reflect underlying /T/. And of course, onset /p/ can occur in all environments.

Although it surfaces in far more complicated ways having to do with fortition and palatalization of following consonants, at a deeper level there can also be said to be a palatal coda /c/. Leaving aside what this is actually realized as, for the most part it behaves like the palatal nasal coda:

Palatal: /ic ɛc ʌc ɔc ac/

Like the palatal nasal rimes, these are neutral to rounding.

What happens with all-neutral lexemes?

While there are some exceptions, in general the following pattern holds true: A lexeme that contains only neutral segments (no oral velars, back vowels, or strong codas) will trigger unroundedness, unless it contains a labial consonant, in which case it triggers roundedness. Thus -tel- + -sU- > -telsư-, but -tebe- + -sU- > -tebesu-.

Exceptions to harmony

Although due to the language's highly inflected nature the harmony system is still quite productive and can be seen in effect in most words, there are some ways in which it doesn't apply. Loan lexemes often have harmony-breaking combinations of segments (especially non-harmonic strong codas), and a small number of morphemes specifically ignore harmony, like Turkish -iyor-. The presence of multiple lexical morphemes in a word (a common occurrence) also "resets" harmony, with the state of inflections being determined by the most recent preceding lexical morpheme.

How did this glorious mess happen?

The current situation developed largely as a result of vowels being colored by neighboring consonants at an earlier phase in the language's history. The glide consonants, essentially already being vowel colorings, were easily swept into harmonizing. Similarly, the dorsal stops /c k kʷ/ had (and still have, to some extent) a slight /i ɯ u/ offglide before a following vowel that wasn't, respectively, /i/, /ɯ/, or /u/, and harmony acted on this to control their distribution.

For a period in Luworese's development, what are currently coda nasals were realized as nasal vowel morae:

/ɲ/ < /ĩ/
/ŋ/ < /ɯ̃/
/m/ < /ũ/
/N/ < /ə̃/, or a nasalized continuation of the preceding vowel's quality (evidence for /ə̃/ mainly being that the present language does distinguish /iN ɯN uN/ from /iɲ ɯŋ um/ in some places that can't be readily explained by analogical leveling)

These mimicked the distribution of their oral vowel counterparts, leading to the distribution nasal codas have in the present language.

Similarly, coda stops appear to have been realized as colored glottal stops at some point in the past:
/c/ < /i̯ʔ/
/k/ < /ɯ̯ʔ/
/p/ < /u̯ʔ/
/T/ < /ʔ/ with no coloring

As with the coda nasals, this vowel coloring came to follow the distribution of the /i/, /ɯ/, /u/, and vowel lengthening components of dimoraic vowels, leading to the current pattern.

The nasal and glottal termini of dimoraic syllables at this stage do derive from earlier nasal and oral stop codas, but they aren't always the same consonant that shows up in the modern language. (Diphthongs before a coda consonant were possible earlier on, for example, leading to different codas today, and alveolar and glottal codas were distinguished.)

In the next post, we'll start looking at how the giant assortment of clusters formed by colliding morphemes actually resolve in speech.

So what do you guys think? Does this make sense?
Last edited by Chengjiang on Wed Oct 19, 2016 3:02 pm, edited 1 time in total.
[ʈʂʰɤŋtɕjɑŋ], or whatever you can comfortably pronounce that's close to that

Formerly known as Primordial Soup

Supporter of use of [ȶ ȡ ȵ ȴ] in transcription

It is pitch black. You are likely to be eaten by a 青.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: Luworese rounding harmony and other morphophonology

Post by WeepingElf »

This is interesting and seems to make good sense to me. I like such things. Rock'n'roll!
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
Chengjiang
Avisaru
Avisaru
Posts: 437
Joined: Sun Feb 26, 2006 4:41 am
Location: Davis, CA

Re: Luworese rounding harmony and other morphophonology

Post by Chengjiang »

Basic consonant cluster resolution

While strictly speaking this question concerns phonotactics, I think it works best to use it as context starting off a section on how segments interact across morpheme boundaries.

As noted, outside of very abstract analyses Luworese can be said to have the syllable structure (C)V(V/F) with a group of finals F consisting of /m N ɲ ŋ p T k s h l/, possibly including /c/. All legal clusters can be straightforwardly derived from an FC sequence. What follow are some rules for generating them outside of certain morpheme-specific situations. Note that while looking at any given word in a vacuum there are no clusters that contain one of /B D/, I've included them here because there are morphemes that can start with one of these consonants and this is how they interact with final consonants from other morphemes. Note also that the presence of underlying final /c/ is determined almost entirely by comparing different realizations of the same final consonant with different following consonants; the only cluster that can be determined to contain a final /c/ without such a comparison is [cːʎ̥], and even then only if the palatal laterals are not regarded as phonemic.

F = /m/:
  • Following glides are converted to the "corresponding" nasal: /mj mw/ > [mɲ mː] (note that due to harmony /mɰ/ can't occur)
  • /mʔ/ > [mp]
  • Following weak consonants are converted to the corresponding nasal: /mB mD/ > [mː mn]
  • All other combinations do not change either segment
F = /N/:
  • /N/ assimilates in POA to the following segment before anything else happens, with glottals and pauses triggering [n]
  • /Nʔ/ > [nt]
  • Sequences of /N/ plus a glide become geminates of the nasal corresponding to that glide: /Nj Nɰ Nw/ > [ɲː ŋː mː]
  • /N/ becomes /l/ before laterals, and any further changes follow from that
F = /ɲ/:
  • These following consonants are palatalized: /n ŋ t k kʷ ʔ ts tɬ s ɬ h l/ > [ɲ ɲ tʃ c c c tʃ cʎ̥ ʃ ʎ̥ ʃ ʎ]
  • Following glides are converted to lengthening: /ɲj ɲɰ ɲw/ > [ɲː]
  • Following weak consonants are converted to the corresponding nasal, with subsequent palatalization of /D/: /ɲB ɲD/ > [ɲm ɲː]
  • All other combinations do not change either segment
F = /ŋ/:
  • Following glides are converted to the corresponding nasal: /ŋj ŋɰ/ > [ŋɲ ŋː] (note that due to harmony /ŋw/ can't occur)
  • /ŋʔ/ > [ŋk]
  • Following weak consonants are converted to the corresponding nasal: /ŋB ŋD/ > [ŋm ŋn]
  • All other combinations do not change either segment
F = /p/:
  • Following glottal consonants become length: /pʔ ph/ > [pː]
  • Nasalizes to [m] before a following nasal: /pm pn pɲ pŋ/ > [mː mn mɲ mŋ]
  • Following weak consonants become stops: /pB pD/ > [pː pt]
  • Following glides fortite to stops: /pj pw/ > [pc pkʷ] (note that due to harmony /pɰ/ can't occur)
  • All other combinations do not change either segment
F = /T/:
  • /T/ > [ʔ] word-finally
  • Following glottal consonants become length with /T/ treated as /t/: /Tʔ Th/ > [tː]
  • Can be treated as becoming a stop assimilated to the POA of any following consonant before any further changes occur
  • Nasalizes before a following nasal, thus producing a geminate: /Tm Tn Tɲ Tŋ/ > [mː nː ɲː ŋː]
  • Non-glottal stops and affricates thus similarly become geminates, with /B D/ merging with /p t/
  • Coronal fricatives become geminate affricates: /Ts Tɬ Tʃ/ > [tːs tːɬ tːʃ]
  • /Tɸ/ > [pɸ], as per above rules
  • Approximants fortite as follows: /Tl Tj Tɰ Tw/ > [tːɬ tːʃ kː kːʷ] (note that Tj does not generate the expected [cː]!)
F = /c/:
  • /c/ vocalizes to a close front vowel word-finally
  • Following glottal consonants become length: /cʔ ch/ > [cː]
  • /B D/ merge with /p t/ before any further changes are applied
  • Nasalizes to [ɲ] before a following nasal, with behavior mimicking /ɲ/: /cm cn cɲ cŋ/ > [ɲm ɲː ɲː ɲː]
  • These following consonants are palatalized: /n ŋ p t k kʷ ʔ ts tɬ ɸ s ɬ h l/ > [ɲ ɲ c tʃ c c c tʃ cʎ̥ ʃ ʃ ʎ̥ c ʎ] (note that, unlike /ɲ/, /c/ also palatalizes /p/ and /ɸ/, and due to the above rule /h/ does not become [ʃ])
  • Following glides become length: /cj cɰ cw/ > [cː]
  • All further changes proceed as if it were /T/ (note that /ctɬ cɬ cl/ > /Tcʎ̥ Tʎ̥ Tʎ/ > [cːʎ̥])
F = /k/:
  • Following glottal consonants become length: /kʔ kh/ > [kː]
  • Nasalizes to [ŋ] before a following nasal: /km kn kɲ kŋ/ > [ŋm ŋn ŋɲ ŋː]
  • Following weak consonants become stops: /kB kD/ > [kp kt]
  • Following glides cause assimilation to a geminate stop: /kj kɰ/ > [cː kː] (note that due to harmony /kw/ can't occur)
  • All other combinations do not change either segment
F = /s/:
  • Following glottal consonants become length: /sʔ sh/ > [sː]
  • Following weak consonants become stops: /sB sD/ > [sp st]
  • Lateralizes to [ɬ] before lateral consonants, with /l/ additionally devoicing: /stɬ sɬ sl/ > [ɬtɬ ɬː ɬː]
  • Palatalizes to [ʃ] before palatoalveolar or palatal consonants, with /j/ additionally assimilating afterward: /sɲ sc stʃ sʃ sj/ > [ʃɲ ʃc ʃtʃ ʃː ʃː]
  • Following velar glides fortite to stops: /sɰ sw/ > [sk skʷ]
  • /sɸ/ > [sp]
  • All other combinations do not change either segment
F = /h/:
  • /hʔ/ > [hː]
  • Following weak consonants become stops: /hB hD/ > [hp ht]
  • Assimilates to a following fricative: /hɸ hs hɬ hʃ/ > [ɸː sː ɬː ʃː]
  • /hl/ > [ɬː]
  • /hj/ > [ʃː]
  • Following velar glides fortite to stops: /hɰ hw/ > [hk hkʷ]
  • All other combinations do not change either segment
F = /l/:
  • /lʔ/ > [lː]
  • /lB lD/ > [lː]
  • Centralizes to [ɹ] word-finally and before central alveolar consonants
  • Palatalizes to [ʎ] before palatals
  • Vocalizes to a close front vowel before palatoalveolar consonants, as a sort of intersection of the above centralization and palatalization rules
  • /lɬ/ > [ɬː]
  • /lj/ > [ʎː]
  • /lɰ lw/ > [lː]
As you can see, there are far fewer distinct consonant clusters than the raw combination of codas with onsets would suggest. Also, under normal circumstances these cluster resolutions are the only way the palatal laterals appear. A single morpheme, -lyO- "together", seems to have a "true" palatal lateral that can't be analyzed as a cluster synchronically; its /ʎ/ interacts with preceding consonants analogously to /l/.

The above is a general account of what consonant clusters end up looking like, whether or not they cross a morpheme boundary. Next, we'll look at phenomena that are specific to morpheme boundaries, such as how word formation converts such things as vowel-vowel hiatus and three-mora syllables into something allowed by the phonotactics.
Last edited by Chengjiang on Fri Oct 21, 2016 5:49 pm, edited 4 times in total.
[ʈʂʰɤŋtɕjɑŋ], or whatever you can comfortably pronounce that's close to that

Formerly known as Primordial Soup

Supporter of use of [ȶ ȡ ȵ ȴ] in transcription

It is pitch black. You are likely to be eaten by a 青.

User avatar
Chengjiang
Avisaru
Avisaru
Posts: 437
Joined: Sun Feb 26, 2006 4:41 am
Location: Davis, CA

Re: Luworese rounding harmony and other morphophonology

Post by Chengjiang »

Notes on orthography

I realized there were some details of the orthography I never explicitly addressed in the opening post. I actually have a question concerning the best way to treat certain sequences in the orthography.

First, there's [ɲ]. I neglected to mention it above, but when it's in a coda (and not part of a geminate), this sound's digraph is reversed from ny to yn: [ɔɲmɔ] is oynmo, not *onymo. This is to avoid it looking to readers like there's another vowel there.

Then, there are some issues with the nasals in general. My current plan is to romanize coda nasals as if they're /N/ whenever possible, with /N/ being m before labials and n everywhere else. This means that for example, /mː nː ɲː ŋː/ are written mm nn nny nng, and /mp nt ɲc ŋk/ are mp nt nky nk. This also means that the postalveolar nasal found when /N/ precedes a palatoalveolar is written n. I think this romanization is the neatest option, and since [n] can't occur before non-alveolar segments it's not ambiguous as to pronunciation. The other possibility I've been considering is to distinguish between strong nasals and /N/ in writing even when they'd be homophonous, by always writing /N/ as n and strong /m ɲ ŋ/ as m yn ng. This means that, for example, /mː ɲː ŋː mp ɲc ŋk/ would be written nm nny nng np nky nk if they contain /N/ and mm ynny ngng mp ynky ngk if they contain strong nasals. (In this system I'd probably also double all letters of other digraphs as well, thus shsh hlhl for /ʃː ɬː/.) This option would make it more obvious what form certain morphemes should take in other words that use them, but I must admit I find it much uglier and less concise, and I think it would tend to lead to spelling pronunciations like the novice's [nm] for nm in romanized Japanese. Also, I'm not fond of making distinctions in writing that don't exist in speech. (If you're wondering, I haven't decided what the native writing system does in these situations, although since none of the nasals are digraphs in it, it's not particularly analogous anyway.)

The same principle as with the nasals goes for other situations where different sets of underlying sounds end up sounding the same. Maybe the sequence [mn] comes from underlying /mn/ in one word and /pn/ in another, but it will be written mn either way.

And then there's /l/. Although it becomes [ʎ] before palatal consonants, since there's no contrast between [l] and [ʎ] there it's still written l. I'm not doing the same thing with [ʃ] before a consonant, though; I think for people who distinguish [s] and [ʃ] at all it would be less intuitive to write it s in that environment. (Except possibly speakers of European Portuguese.)

That all said, what do you think concerning the nasal codas and other cases where two different sequences have the same surface realization? Should I stick with following pronunciation, or show the strong/weak distinction and other such distinctions even when they're homophonous?
[ʈʂʰɤŋtɕjɑŋ], or whatever you can comfortably pronounce that's close to that

Formerly known as Primordial Soup

Supporter of use of [ȶ ȡ ȵ ȴ] in transcription

It is pitch black. You are likely to be eaten by a 青.

User avatar
Chengjiang
Avisaru
Avisaru
Posts: 437
Joined: Sun Feb 26, 2006 4:41 am
Location: Davis, CA

Re: Luworese rounding harmony and other morphophonology

Post by Chengjiang »

OK, resuming this thread with a bit of a change of plans. I want the current version of Luworese to have Finnic-style consonant gradation. The original did not. I'm having a bit of trouble setting up something that looks good since both the consonant inventory and the allowed clusters are more complex in Luworese. These are the correspondences I'm sure of thus far:

Strong :: Weak
p :: b
t :: d
ky :: y
k :: g
kw :: w
mp :: mm
nt :: nn
nky :: nny
nk :: nng
nkw :: mm, sometimes nng
lp :: ll
lt :: ll
lky :: lly
lk :: ll
lkw :: ll
CC :: C, e.g. mm :: m, kk :: k, or ss :: s
Probably also heterorganic nasal + stop clusters participate with the stop weakening and the cluster resolving as per above, e.g. ngt :: ngn or mky :: mny

As you may notice, these are very similar to the gradations found in Finnish. However, I didn't feel like stopping there. In particular, I wanted to work the glottal stop into the system as a possible weak grade for some short sonorants, as a kind of systematic version of Hawaiian's sporadic reduction of various consonants (besides universal reduction of earlier Polynesian *k) to the glottal stop. I figured it might evolve from them reducing to zero and the resulting hiatus generating a glottal stop. How does this look?

Strong :: Weak
n :: ' (not sure if I want to include the other nasals)
l :: '
h :: '

(Yes, I know [h] isn't typically considered a sonorant, but it does reduce to [ʔ] here.)

Although the affricates and the sibilant fricatives don't have distinct strong and weak grades (other than long :: short), I was also considering these:

Strong :: Weak
f :: b
hl :: l

Consonants and clusters other than those mentioned above do not participate.

Does this all make sense? Should I change anything?
[ʈʂʰɤŋtɕjɑŋ], or whatever you can comfortably pronounce that's close to that

Formerly known as Primordial Soup

Supporter of use of [ȶ ȡ ȵ ȴ] in transcription

It is pitch black. You are likely to be eaten by a 青.

Post Reply