An, Young-ran. 2016. Capturing variation and gradience in identity avoidance: A case of machine learning. Studies in Phonetics, Phonology and Morphology 22.3, 533-557.
This paper makes the point that a grammar appears to be a sum of tendencies, rather than an aggregate of all-or-none instances. That is, the grammar is not formed by an across-the-board law, but teems with variation and gradience. For a case in point, this paper presents the phenomenon of consonant insertion in Korean total reduplication. To see whether this kind of grammar with variation and gradience can be possibly, and eventually humanly, learned, it is simulated using a model of grammar learning. The instantiation of machine learning in this paper shows that a grammar with variation and gradience can indeed be learned. (KC University)