Cyran - Between phonology and phonetics: Polish voicing

The original aim of this study was to provide a comprehensive phonological description of the Polish voicing complex within a minimalist model of phonology, which can be briefly characterized as a representation-based approach, with strict privativity of melodic representations, and a highly constrained view on the nature of possible phonological processes. The theoretical framework that was selected for this purpose is a version of Government Phonology (Charette 1991; Cyran 2010; Gussmann 2007; Harris 1994; Kaye, Lowenstamm and Vergnaud 1990; Lowenstamm 1996; Scheer 2004).

The main reasons for our dissatisfaction with the previous analyses are two-fold. Firstly, some aspects of Polish voicing, such as Cracow-Pozna sandhi and Progressive Voice Assimilation, have not received explanatory accounts, especially within the representational and privative tradition. Secondly, the accounts that seem to work are couched in theoretical frameworks which allow for too much in terms of both representational and computational machinery. For example, we find that such artefacts as binary features, specification of sonorants, extrinsic rule ordering, and some rules that have been proposed, as well as the direct and indiscriminate translation of phonetically observed patterns into a formal phonological description are hardly illuminating.

Given that the research programme which can be generally called Laryngeal Realism (e.g., Gussmann 2007; Harris 1994; Honeybone 2002; Iverson and Salmons 2003b) which is closest to our views on phonological organization also fails to provide satisfactory answers concerning CP sandhi and PVA, it was necessary to propose and apply a different analytical tool: Laryngeal Relativism. The main idea is that sound patterns stem from the structure of Sound Systems which consist of three independent aspects: phonology, phonetic interpretation and phonetics. The phonological component includes phonological representation and computation. The result of computation is still a phonological representation which is subject to systemic interpretation. We claim that the relation between phonological categories and phonetic categories is arbitrary. This in turn leads to the claim that phonology can be substance-free, but the Sound Systems are not, because they involve the connections (phonetic interpretation) which have been established in language acquisition.

Throughout this book we tried to demonstrate that sound patterns may have a phonological, interpretational or purely phonetic basis, and that the exact division must be established to understand the phenomena in question. As a result of arbitrariness between phonological and phonetic categories, Polish may be analysed as a language in which two opposite Laryngeal Systems are used: an H-system in Cracow-Pozna Polish and an L-system in Warsaw Polish. It was shown that all commonly known voicing phenomena can be handled by the reversed systems, except that the same phonetic effects must now be viewed as following from different aspects of the Laryngeal System in question. For example, FOD is phonological in WP, and interpretational in CP. Sandhi voicing assimilation shows the same distinction. It can only be phonological in WP, and must be purely phonetic in CP. All this leads to a situation in which nothing is immediately obvious from the point of view of the system.

Laryngeal Relativism forces us to reconsider practically all aspects of analytical methodology, especially the status of phonetic facts. A similar point can be made about the carrier signal modulation theory (Harris 2009; Ohala 1992), in that the presence of displacement from a carrier signal does not immediately mean that it is linguistically relevant, unless what we mean by linguistic overlaps with the definition of Sound System, rather than phonology alone. In this sense, Laryngeal Relativism is compatible with the carrier signal modulation theory. As in syntax, we must allow for the possibility that the acoustic signal is ambiguous from the linguists point of view. Except that instead of structural ambiguity, we are dealing with the ambiguity of phonetic interpretation conventions which link phonetic categories to phonological representation.

As for voicing phenomena and phonological analysis, we have seen that the basic criteria for phonological marking of an obstruent as voiced that is, full voicing and the presence of voicing assimilation may be wrong. Similar problems concern well-established concepts such as sonorant, obstruent, obstruentization, and especially the equating of the phonetic presence of friction with obstruency.

The structure of the Sound System used in this book makes certain suggestions also about the nature of laboratory phonology research. The central question of that research is how acoustic signal and auditory modelling can be used to judge which kind of data constitutes the empirical basis of phonological analysis (e.g., Harris 2009; Jessen 2001). The perspective imposed on such research by the findings in this book is that the kind of evidence that has normally been taken to indicate the type of phonological system we are dealing with is in fact the evidence of how the entire system works, but it is wrong to assume that system equals phonology. The latter constitutes the bare minimum of the system. The question is, rather, what kind of evidence can be used to determine both the phonological side of the sound system (representation and computation) and the phonetic interpretation conventions which connect this abstract component with phonetic facts. In other words, the phonetic facts are not a direct reflection of the phonology of a given sound pattern.

It seems, that in this point, we agree with Keating (1984). Keating uses the features [voice] and [+voice] to represent the phonological two-way laryngeal contrast in both voicing and aspiration languages, for reasons of universality of representation and rule equivalence across languages, leaving the actual phonetic implementation to language specific rules. The main difference between Keating (1984) and this study lies in the fact that we use a strict privative system of representation, which generates the difference in the interpretation of the non-specified obstruents depending on whether they are part of an H-system or L-system.

More research is needed into the nature of licensing. We saw that licensing is directly responsible for the distribution of laryngeal categories. The connection between Laryngeal licensing and pragmatics, which transpired from the discussion of FEN strengthening effects requires more research too. It appears that licensing strength is manipulated at the interface between phonology and pragmatics.

It is hoped that this book will contribute to the research on linguistic interfaces, especially to the turbulent relationship between phonology and phonetics.