Vocalic Roots

So, I'm attempting to conceive of a proto-lang that is completely nuts. Basically, I want to start somewhere very unnatural and make it as natural as I can. Semitic language begin with tri-consonantal roots. What if some kind of tri-vocalic root system arose? What would the inherent properties of the vowels be? What would left open for completion in the verbal and nominal systems? The easy way would've been to just come up with more than 20 vowels (plain, nasal, long, unrounded, etc.) and less than 10 consonants, to parallel the 30ish consonants and 10ish vowels of Proto-Semitic. But taking a step back towards the naturalistic, I saw vowels don't behave like that. Vowels affect thing around them (assuming they're in charge) or they get affected by the consonants around them. So I looked into vowel harmony, and I thought, "I should have consonant harmony" and not just of the mutational variety either. The next big decision is to make everything voiced. Most likely, this is because my biggest conlang is all unvoiced, but I also think the language is pro-singing, i.e. pro-voiced. The four biggest categories for consonants that we would want vowels to spawn are

• nasals
• stops
• fricatives
• approximants

'''Nasalization''' is easy enough to understand. (Pre-)'''Glottalization''' might explain stops. Approximants are often co-articulated, e.g. /j/ is just palatalization, /w/ is labialization, etc. '''Fricitivization''' is not a word, but Chinese linguists were so eager to add ɿ, ʅ, ʮ, and ʯ that there must at least be some linguistics who think it ought to be. /i/ and /j/ are related, as are /u/ and /w/. It is disputed, but rhotic approximants come from rhotacized vowels, and /e/ might be related to /l/. I am pretty good at a Chewbacca impression, and I get it by co-articulating everything with a uvular trill/approximant. This can be done with any other voiced approximant and will all the vowels (though /i/ is most difficult). Lastly, some consonants can only be made by getting the tongue all the way out of the mouth. This could be called (and I'm totally making this up) '''Advanced Tongue Tip'''. This could be thought of as having two degrees, one dental and one alveolar.

	Nasalization	Glot.	Fric.	Labial.	Uvuv.	Velar.	++ATT	+ATT	+RTR	Palat.
Glottalization	No
Fricitivization	No	No
Labialization a.k.a. Rounding	m	b	v
Uvularization	ɴ	ɢ	ʁ	vʶ
Pharyngealization a.k.a Velarization	ŋ	g	ɣ	w	ɣʶ
++ATT	n̪	d̪	ð	Yes	ðʶ	ɫ̪
ATT	n	d	z	Yes	zʶ	ɫ
RTR	h	ʔ	ž	Yes	ř	Yes
Palatalization	mʲ	bʲ	vʲ	Yes	ɴʲ	ŋʲ	ʎ	nʲ	hʲ
Ø	No	No	No	Yes	ʀ	ɰ	Yes	Yes	ɻ	j

"Yes" means a consonant is underspecified. Considering only RTR for a minute, here are the basic vowels (ATR on the left, RTR on the right):

	Front	Mid	Back
High	i ↔ ɪ		ʊ ↔ u
Mid	e ↔ ɛ	ə*	o ↔ ɔ
Low	æ ↔ a		ɐ ↔ ɑ

Nasalization and rounding will be considered separately. Rhotacized will be the same as "schwa-ified".
Finally, a Ph.D dissertation I found delineates articulation thusly:

1. Front
   1. Suction
   2. Continuant
   3. Strident
   4. Lateral
2. Labial
   1. Round
   2. Labial
3. Coronal
   1. Anterior
   2. Distributed
   3. Coronal
4. Dorsal
   1. High
   2. Low
   3. Back
   4. Dorsal
5. Back
   1. Nasal
   2. ATR/RTR
   3. Radical
   4. laryngeal (creaky voice)