Phonetics-phonology interactions in pre-sonorant voicing
Patrycja Strycharczuk
November 2012

The present dissertation is a laboratory phonology exploration of obstruent voicing before a following sonorant sound. Pre-sonorant voicing shows a number of recurrent typological properties that are challenging from a phonological point of view. First, voicing applies in the environment of a segment that is not contrastively voiced. This raises the question whether sonorants may be redundantly specified for laryngeal features, and, if so, how such specifications may be assigned. Second, voicing is positionally restricted to environments where laryngeal neutralisation is also observed, which has led some phonologists to propose a conditioning link between the two processes. Third, pre-sonorant voicing displays an array of manner asymmetries, with fricatives being more prone to undergo the process than stops, even though there is no clear evidence for a preference for voiced fricatives over voiced stops in the languages of the world. The segmental and positional asymmetries involved in conditioning pre-sonorant voicing coupled with questions concerning the role of phonological contrast have been the subject of active debate in theoretical phonology. The present thesis attempts to illuminate this debate through systematic analysis of acoustic data from four languages: West-Flemish, Poznan Polish, Central Catalan and Quito Spanish. The primary research question driving the exploration of the results is whether the studied cases of pre-sonorant voicing show evidence of phonetic gradience or whether their phonetic behaviour is more consistent with a categorical interpretation. Building up on this, I consider which aspects of pre-sonorant voicing may be phonetically conditioned, and what level of phonological abstraction is required to accommodate the observed empirical facts. Empirical diagnostics for categoricity and gradience point towards the existence of cases of categorical pre-sonorant voicing which call for a phonological analysis. However, there is little evidence to support the hypothesis that the formal representation of pre-sonorant voicing involves redundant laryngeal specifications for sonorants. Instead, I propose that obstruents subject to pre-sonorant voicing acquire a laryngeal specification not via feature spreading, but through a context-sensitive process of default feature assignment targetting delaryngealised sonorants. This process arises diachronically from listeners misperceiving passive voicing as the realisation of a voicing target. As previously proposed by Jansen (2004), delaryngealisation facilitates extended passive voicing by removing active devoicing gestures. Passive voicing may then be reanalysed as intentional. I extend this perceptual explanation to account for manner asymmetries in pre-sonorant voicing. The argument draws on perceptual asymmetries concerning acoustic cues to voicing in stops and fricatives, which are potentially more salient in passively voiced fricatives than in passively voiced stops. While pre-sonorant voicing reflects some influences of functional pressures shaping its diachronic development, the resulting voicing patterns are not always synchronically functional. For instance, voicing in stop+sibilant clusters but not in singleton stops, as seen in Catalan, is unlikely to reflect the direct phonologisation of phonetic pressures. Rather, I propose a diachronic scenario involving rule telescoping: a pattern of intervocalic sibilant voicing is first reanalysed as prevocalic by rule generalisation, and the voicing target assigned to the prevocalic sibilant later triggers anticipatory assimilation. The role of abstract phonology, manifested here through rule generalisation, can be further observed in the distribution of pre-sonorant voicing, which in some cases is sensitive to the boundaries of abstract morphosyntactic domains, as instantiated by the cases of Catalan and Quito Spanish. I propose that functional phonetic and abstract phonological pressures on pre-sonorant voicing are best reconciled in a diachronic perspective, where phonetic factors play a crucial role in the initial stages of sound change. These phonetic factors are understood as influences on language use, rather than being cognitively represented in the speakers' grammars. The primary role of phonology involves replicating the patterns present in the input without direct influence from phonetic factors. This idea involves a vision of phonology which is relatively abstract, but which nevertheless retains a transparent mapping to the phonetics. Pre-sonorant voicing data lend empirical support to such a conception of phonology by showing evidence of categorical phonetic behaviour which does not correspond to phonological categories defined top-down using the criterion on contrast. Based on these cases, I argue for a bottom-up model where abstract features emerge from the continuous phonetics, and where categoricity, rather than contrast, is the criterion that determines whether or not a distinction is represented in the phonology.
Format: [ pdf ]
Reference: lingbuzz/001645
(please use that when you cite this article)
Published in: University of Manchester
keywords: phonology; phonetics; pre-sonorant voicing; dutch; catalan; polish; spanish, phonology
Downloaded:1494 times


[ edit this article | back to article list ]