Sunday, July 10, 2016

Detail #299: An Idea for a Morphophonology

So, I have been planning this morphophonological system for Sargaĺk, originally, but have abandoned it because it just didn't really fit with the imaginary history of the language.

The idea is that morphemes do not consist of strings of phonemes (although phonemes are also permitted to occur), but by strings of underspecified phonemes. Thus, you can have a thing like
passive: {+back}l{-occlusive +voiced}
Depending on the phonological context in which this appears, phonemes will be picked to realize these features using some set of rules. The fun appears when multiple morphemes appear together. Also, "weak brackets" would be ones that permit for merging with the previous/next phoneme. A hierarchy of features would determine how the resulting set of features ends up. Let us use {/ and /} for such brackets.

Each underspecified phoneme could of course maybe have a 'strongest' feature that could be written in bolds or as the leftmost feature or whatever, to which special rules apply. Coming back to the passive marker presented previously, we could suffix the following morpheme:
first person: {/ +semivowel}a
Thus we get 
{+back}l{-occlusive +voiced}{/ +semivowel}a → {+back}lwa
What maybe happens there is that the backness present before l velarizes l, which leads to +semivowel coming out as /w/ rather than /j/. 

Of course, at times a combination of features just might not be realizeable, say e.g. {+occlusive/} {/ -occlusive} or {+velar /} {/ -velar }. For such conflicts, a set of rules need to exist. At other times, the interpretation need not be entirely literal.

No comments:

Post a Comment