Let's suppose we have a language that underlyingly only distinguishes length and tone. With two degrees of length for vowels and consonants plus five tones, we arrive at 2*2*5=20 possible CV syllables. If we also allow for V and CVC syllables, this yields ((2*5)+(2*2*5)+(2*2*2*5))^2=4900 possible two syllable words.
(1) Underlying Inventories
/C Cː/ <C CC>
/V Vː/ <V VV>
/˥ ˧ ˩ ˥˩ ˩˥/ or /V́ V̄ V̀ V̂ V̌/ <V́ V V̀ V̂ V̌>
Additionally suppose that the features for Place of Articulation and Manner of articulation are autosegmental, i.e. for each word there is a sequence or meldody of (maximally three) feature values for for these features for a word. These are associated from left to right. This yields the following vowel and consonant segments in (2) and (3). Notice that the vowel height feature and the manner of articulation for consonants are the same feature.
(2) Consonant Segment Inventory
Code: Select all
Labial Coronal Dorsal
p t k closed <p t k>
f s x half closed <f s h>
w l ʁ open <w l r>
Code: Select all
Labial Coronal Dorsal
u i ɨ closed <u i y>
o e ə half closed <o e a>
ɔ ɛ a open <ọ ẹ ạ>
(3) Example word
/CV́ːCːV̂/+Place:labial-coronal-labial+Manner:closed-open-half closed --> pɛ́ːfːô <pẹ́ẹ́ffô> 'dog'
How is this different from a normal segmental inventory and assimilation of vowels and consonants outside a certain domain. If morphology comes into play, we can get some interesting results. Let's first talk about material that could be underlying, e.g. a /V̌/ prefix. We have to distinguish two possibilities, early affixes and late affixes. Let's talk about early affixes first. If we affix the above mentioned prefix before assigning any of the manner and place features, we get a very differnt word. Let's say that this prefix expresses plurality. We thus get a seemingly suppletive plural for dogs in (4).
(4) Example word with early /V̌/-prefix
(/V̌/+/CV́ːCːV̂/)+Place:labial-coronal-labial+Manner:closed-open-half closed --> ǔlóːfːô <ǔlóóffô> 'dogs'
If the /V̌/ prefix attaches after manner and place assignment it will just receive the manner and place features of the first segment. In our case this means it will surface as ǔ <ǔ> thereby forming a minimal pair with (4). The form of the stem does not change here. Let's say this prefix indicates a first person possessor in (5).
(5) Example word with late /V̌/-prefix
/V̌/+(/CV́ːCːV̂/+Place:labial-coronal-labial+Manner:closed-open-half closed) --> ǔpɛ́ːfːô <ǔpẹ́ẹ́ffô> 'my dog'
Additionally, we could assume that melodic, i.e. non-underlying features, are affixed. Again we have to distinguish two cases. The affixes can either be overwriting or additive. Let's start with an additive suffix -dorsal. This suffix will change any segment into a dorsal that would otherwise be assigned its place feature by spreading from a previous segment. For our example word it means that the last ô <ô> is changed into ə <a>, as seen in (6). I assume that this is a locative exponent. While this looks like usual vowel mutation it will affect longer chunks. This is true as long as these get their place and manner values by spreading. It thus affects the last vowel and the last consonant in (7)
(6) Example word with -labial suffix
/CV́ːCːV̂/+Place:labial-coronal-labial+Manner:closed-open-half closed+Place:dorsal --> pɛ́ːfːô <pẹ́ẹ́ffâ> 'at a dog'
(7) Example word with with early /V̌/-prefix and -labial suffix
(/V̌/+/CV́ːCːV̂/)+Place:labial-coronal-labial+Manner:closed-open-half closed+Place:dorsal --> ǔlóːʁːâ <ǔlóórrâ> 'at dogs'
The other case for melodic affixes would be overwriting. Let's assume we have a melody for manner open-closed that expressed definitness. This will overwrite the lexical closed-open-half closed melody of the word dog. The word for 'the dogs' again looks suppletive, even though it is just a regular case of overwriting affixation.
(3) Example word
/CV́ːCːV̂/+Place:labial-coronal-labial+
I hope this is possible to understand my ideas at least to some extent. If not, you can ask questions, of course. I am happy to answer them.