Polish palatalisations: the role of frequency and phonological naturalness

The paper focuses the status of Polish palatalizations and argues that they should be reanalyzed as morpheme-specific consonantal mutations. Evidence is presented indicating that they cannot be viewed as resulting from phonologically natural operations. The applicability of the processes appears to be due to morphological and lexical factors, i.e. the presence of a particular morpheme in a particular word. Frequency of use comes out as more efficient in predicting the consonantal mutations than markedness. First, in the case of two patterns that compete in one morphological context, the one that is more frequent is productive. Second, morphophonological constructions that show a higher frequency of use are more stable than those of a lower frequency. The latter are more susceptible to modifications caused by the preference for transparent bases.