Zompist Bboard Again

I've crossposted this topic for the Runepath roleplay setting's main language, the name of which loosely translated as Common, on the CBB. After an experience with a similar topic, in which I dumped everything on them at once, I'm posting one stage at a time, starting with the protolang.

Is the bit about [*ɜ] deleting below phonotactical or morphophonological?

kcüNqaulnêlthöïhë [k̩sʊ.ŋɑ̃ũ̯l.ɖɜlˈtʰɔɪ̯.ħɛ] (Proto-Common)
Phonology
Simple consonants
/p pʰ t tʰ k kʰ ʔ/ <p ph t th k kh q>
/p͡r̥ t͡r̥ k͡r̥/ <pr tr kr>
/s ʝ ħ/ <c j h>
/m n ŋ/ <m n ñ>
/ɾ ʎ̥/ <r l>

Consonant clusters
[pʝ tʝ ks kʝ mʔ nʔ ŋʔ] <pj tj kc kj mq nq ñq>

Vowels
/æ ɑ e ɛ ɜ i ɪ o ɔ u ʊ y/ <a ä e ë ê i ï o ö u ü y>
/ɑɪ̯ ɑu̯ ɔɪ̯/ <aï au öï>

Allophony
[r] of [ɾ] after [p t k]
[ɽ̊͡r̥] of [ɾ] between [p t k] and [ɜ]
[ʈ ɳ ɽ k͡ɽ̊͡r̥ p͡ɽ̊͡r̥ t͡ɽ̊͡r̥] of [t n ɾ k͡r̥ p͡r̥ t͡r̥] before [ɜ]
[ɑ̃ ɑ͠ɪ̯ ɑ͠ʊ̯ æ̃ ẽ ɛ̃ ĩ ɪ̃ õ ɔ̃ ɔ͠ɪ̯ ũ ʊ̃ ỹ] of [ɑ ɑɪ̯ ɑu̯ æ e ɛ i ɪ o ɔ ɔɪ̯ u ʊ y] after [m n ŋ]

Other notes
[ks pʝ tʝ kʝ mʔ nʔ ŋʔ] are better-analyzed as clusters because vowels can be transfixed between the components
[æ] is IPA [a]

Phonotactics
Free word order with XSV as default
(C₁C₂)V(NC₃) syllables
—all but [r̥ ɽ̊͡r̥] as (C₁), [ʔ s ʝ r̥ ɽ̊͡r̥] as (C₂), [m n ŋ] for (N), [ʔ] as (C₃) only after (N)
[ɜ] deletes adjacent to other vowels. When between [t t͡r̥ p͡r̥ k͡r̥ r] and another vowel, it results in [ʈ p͡ɽ̊͡r̥ ʈ͡ɽ̊͡r̥ k͡ɽ̊͡r̥ ɽ] preceding that vowel.
Stress is on the next-to-last syllable.

Root structure
Most, but not all, words derived roots of two-to-four vowels that include syllable-final [ɾ ʎ̥] in the vowel
Roots can be compounded. For example, the root meaning "ground/earth", ɑu̯l-ɜ, combines with the one relating to language, ɜ-ɔɪ̯-ɛ, to form the one relating to Common, itself—ɑu̯l-ɜl-ɔɪ̯-ɛ.

The language posted above is only one factor is Common's construction because humans, like us, only provided limited influences to the final language. The next two posts will focus on the language's major contributors, Elvish and PrimordialPhoenixfolk. Unfortunately, Elvish may take a while to appear because I'm not gonna spend time fleshing out another language right now as languages are just flavor.

<Primordial redacted due to changes. See next post for Phoenixfolk.>

Feel free to correct my classifications, if you need to, because this is my most complex phonology to date.

Phoenixfolk (Romanization only)
Phonology
Consonants
/p̪ b̪ t d c ɟ k g/ <p b t d kj gj k g>
/p̪ʲ b̪ʲ/ <pj bj>
/p̪ʷ b̪ʷ tʷ dʷ kʷ gʷ/ <pw bw tw dw kw gw>
/ʈ͡ʂ t͡ɕ d͡ʑ c͡ç ɟ͡ʝ/ <qj tj dj kj gj>
/f v s z ʂ ɕ ʑ ç x/ <f v s z q sj zj xj x>
/fʲ vʲ/ <fj vj>
/fʷ vʷ sᶲ zᵝ sʷ zʷ ʂʷ xʷ/ <fw vw s(j) z(j) sw zw qw xw>
/ɱ n ɲ ŋ/ <m n n(g)j ng>
/ɱʲ/ <mj>
/ɱʷ ŋʷ/ <mw n(g)w>
/j ɥ w/ <j jw/wj w>
/ɾ/ <r>
/ɾʲ/ <rj>
/ɾʷ/ <rw>
/N H Q Y/ <n ñ p/b/t/d/k/g a/i/o/u/e> (moraic)

Vowels (parenthetical phonemes have the same orthography)
/(ɑ œ) i (o̞ ø) (ɯ y) ə/ <a i o u e>
/Ṽ/ <Vñ>

Allophony (details below)
[ɱ n ŋ] of [N]
[p̪̚ b̪̚ t̚ d̚ ʈ̚ k̚ g̚ s z l] of [Q]
[ɑ i o̞ ɯ ə] of [Y]
[ɑ̃ œ̃ ĩ õ̞ ø̃ ɯ̃ ỹ ə̃] of [VH]

Complementary distribution
<pj bj fj vj mj> series: [p̪ b̪ f v ɱ] before [œ œ̃ ø ø̃ y ỹ] and [p̪ʲ b̪ʲ fʲ vʲ ɱʲ] before [ə ə̃]
<kj gj> series: [c ɟ] before [ɑ ɑ̃ o̞ õ̞ ɯ ɯ̃ ə ə̃] and [c͡ç ɟ͡ʝ] before [i ĩ]
<s z> series: [​s z] before [ɑ ɑ̃ i ĩ ə ə̃] and [sᶲ zᵝ] before [o̞ õ̞ ɯ ɯ̃]
<sj zj> series: [ɕ ʑ] before [ɑ ɑ̃ i ĩ ə ə̃] and [sᶲ zᵝ] before [ø ø̃ y ỹ]
<wj> series: [w] before [œ œ̃ ø ø̃ y ỹ] and [ɥ] before [i ĩ ə ə̃], where it merges with the <jw> series

Other notes
The script's native calligraphy is called rjuna [ɾʲɯ.nɑ], collectively, with the "printed" form called maqarjuna [ɱɑ.ʂɑꜜ-] and the "cursive" form called vjeñrjuna /vʲə.Hꜜ-/ [vʲə̃ꜜ-]. Unlike its Japanese inspiration, /N/ is velar [ŋ] finally, as original /H/ was [ŋ̊] (more below). The pheonixfolk use this script and the Latin alphabet so Romanization has always been official, as opposed to the multiple Romanization schemes we have for some Asian languages.

Phonotactics
(CG)V morae with special (N H Q Y) permitted after V conditionally
—unrestricted (C), [j w] as (G)
——<GG> onset is acceptable with the other glide; [j] is forbidden between [p̪ b̪ f v ɱ ɲ ɾ] and [i ĩ]
—Special mora Q geminates the following consonant's primary articulation except before [c ɟ c͡ç ɟ͡ʝ f fʲ fʷ v vʲ vʷ ç x xʷ ɱ n ɲ ŋ j ɥ w ɾʲ], where it's forbidden. It yields [​s] before [ʂ ɕ], [z] before [ʑ], and [l] before [ɾ ɾʷ], which become [l lʷ]—in turn. Additionally, special mora N assimilates to the following consonant's POA except before [c ɟ c͡ç ɟ͡ʝ ʂ ɕ ʑ ç ɲ j ɥ], where it's forbidden. Special mora H was N's voiceless counterpart before generating nasal vowels when its pronunciation was lost. Yet, H is preserved in phonemic transcription for purposes of stress (see below)
—special mora Y is forbidden after [œ ø y] because <j> isn't directly in front of it
Vowels are pronounced separately
A downstep occurs after the first special mora or the third-to-last mora. If H had been the first special mora, the downstep occurs after the mora containing the resulting nasal vowel