ON THE THEORETICAL PROBLEMS OF PREDICTABILITY IN SERBIAN PALATALIZATION

The paper presents the Serbian velar palatalization: /k/ → [tʃ], /ɡ/ → [ʒ], /x/ → [ʃ]. A detailed, comprehensive description of the phenomenon will not be provided, as it is not the main task and theme of the paper. The paper demonstrates that predictability does not strictly hold „in equal measure” in phonology, contrary to what traditional grammars suggest (e.g., Klajn 2005). Even theory-specifi approaches do not address it (e.g., Starčević 2005). This is one of the foundational assumptions of the theory of Government Phonology à la Kaye et al. (1985; 1990): phonological processes are organized according to universal principles and are not arbitrary (Non-Arbitrariness Principle). Universal rules can be overridden by parametric settings, which characterize the phonology of a specific language – in this, Government Phonology shares principles parallel to those of the Minimalist Program. The task of the paper is to show when predictability is not met, through the lens of Serbian phonology, based on the aspect that shows I-U asymmetry. The paper’s implicit goal is to highlight that the strict conditions imposed on phonological theories may not necessarily account for the variations shown on the surface, and things are much more complicated than what we might assume about Universal Grammar. The paper follows the theory of Government Phonology along with Element Theory (Backley 2011), which will be essential for explaining the phenomenon.