1 In: Bondaruk, A. & A. Malicka-Kleparska, (eds.) 2013. Ambiguity – Multifaceted structures in syntax, morphology and phonology. Lublin: Wydawnictwo KUL. 271-298 Representational ambiguity in Polish palatalization Geoffrey Schwartz This paper presents an analysis of Polish velar and coronal palatalization processes within the Onset Prominence representational environment (OP). OP builds on recent insights into the structural nature of segmental representation to provide a novel perspective on phonological constituency. In the OP environment, palatalization is a phonotactic phenomenon, sparked by the fusion of consonantal and vocalic structures, and motivated by the need to avoid melodic ambiguities that may arise during this fusion process. In particular, we will note the inherent ambiguity of velar consonants, whose acoustic and perceptual properties reflect the context dependence of sounds articulated with the body of the tongue. The realization of the palatal element {I} among the various levels of the OP hierarchy allows us to unify representational and phonetically-oriented explanations of palatalization. 1 Introduction The relationship between phonetics and phonology constitutes an important area of ambiguity in linguistic structures. The fundamental source of these issues lies within the realm of speech perception, and in particular, the discrepancies between the acoustic signal and how it is parsed by listeners. Stated briefly, the same acoustic input may be mapped to different phonological structures in different languages, as evidenced in numerous experimental studies in cross-language speech perception (e.g. Flege 1995, Best 1995). In light of this problem, a desirable objective for theories of phonological representation is to derive the possible interpretations of the perceptual ambiguities found in speech. This contribution will therefore present a newly developed framework, the Onset Prominence representational environment (Schwartz 2010, 2013), which contains built-in ambiguities of phonetic origin. The empirical focus of this paper will be the complex set of alternations in Polish that are commonly referred to as palatalizations. In modern Polish these alternations present a major challenge for phonological analyses 1 , since they are largely dependent on morphological considerations. For example, the locative suffix -e added to the word kot ‘cat’ yields kocie /kotɕe/, in which the /t/ is mutated oʃ ‘palatalized’ to /tɕ/. Conversely, the diminutive suffix -ek produces the form kotek with no change in the stem-final /t/. 2 Thus, we are faced with a situation in which the vowel /e/ is associated with an alternation in some suffixes but not in others. In the Onset Prominence environment, we shall see that the apparent ambiguous behaviour of front vowels such as /e/ may be described in terms of a phonetically-derived representational ambiguity. In other words, we shall offer a phonological explanation of how a system with two representationally distinct yet phonetically indistinguishable vowels may evolve. Indeed ‘evolve’ is an appʃopʃiate woʃd here, since morphophonological alternations typically entail the diachronic grammaticalization of changes with phonological origins. Enriched synchronic representations allow us to explain the diachronic process by which the complexities associated with palatalization in Polish may emerge. Briefly stated, phonetically-motivated ambiguities in the OP representational system were subject to different interpretations during the various stages in the phonological development of Polish. Note heʃe that the teʃm ‘inteʃpʃetations’ ʃefeʃs fiʃst and foʃemost to the listeneʃ, who is obligated to paʃse the incoming signal. Thus, the mechanism of change is fundamentally listener-induced (Ohala 1981, Blevins 2004), though the changes themselves may have articulatory motivation. That is, even the coarticulation of two sounds is a phenomenon with perceptual consequences. Listeners must decide 1 Due to the fact that this paper is presenting a relatively new and unfamiliar theory, we will devote only minimal attention to the numerous previous analyses of palatalization in Polish. Suffice it to say that our goal is to provide a representational account in which phonetic considerations are captured. Space restrictions prevent a detailed critique of other approaches. However, it may be said that such critiques would be largely based on assumptions about the status of the segment (cf. Gussmann 2007: 45). 2 Even greater complexity is revealed in the case of stem-final dorsal consonants such as /k/ in word sok ‘juice’, whose diminutive is soczek /sotݔek/ (see 4.5).