Epenthesis as a matter of FAITH

In Optimality Theory, the observation that default epenthetic segments are sourced
from a heavily restricted set of segments across languages, such as the glottal stop
[ʔ] for consonants, schwa [ə] or [i] for vowels, is generally analysed as a
 markedness effect: Optimal epenthetic segments are maximally unmarked segments. This
chapter highlights several conceptual and empirical problems with this claim and
proposes a faithfulness-based alternative instead: Optimal epenthetic segments
 involve minimal epenthesis at the level of distinctive features, that is, minimal
violations of Dep(F). We will show how this proposal accounts for crosslinguistically
common epenthetic segments as segments that are in some way underspecified
vis-à-vis other segments, and it will be shown how the problems encountered in a
markedness-based approach can be resolved.