Detail #279: Morphophonology to the Extreme

I bet this is attested in some real-world language, although it might be the case that this kind of analysis hasn't been applied to that language. 

So, my idea is that instead of a morpheme consisting of a set of allomorphs, each consisting of a string of phonemes, one could have morphemes consisting of features. This means instead of, say,

-iŋ

you would have something a bit like

{+vowel, +front, +closed, ...}{+nasal, +velar, +consonant ...}

The main difference, however, would be that some of the elements would underspecify what phoneme to be realized. Instead of -ing or the featural form of -ing, we'd have

{+front, +closed}{+back, -obstruent}

Different combinations of features would have different 'favoured' realizations, but morphophonological context would sometimes lead to a different realization. Sometimes, two different sets of features would be forced to merge, with hierarchical rules resolving what feature wins if there's a conflict - say between +obstruent and -obstruent, for instance.  

Now, if your morphology is no richer than that of English, this wouldn't be very interesting, but once we go hugely agglutinating, this could lead into interesting things where morphemes that underlyingly are very agglutinating, end up looking almost fusional, as features get suppressed, assimilated, meta-phonemes get merged into new phonemes that drop some features, etc.

The verbal morphology of Sargaĺk will be an example of this, but working it out has turned out to be pretty challenging.