Phonological Involvement in Phonetic Change

Josef Fruehwald (2012)

Abstract

A long standing issue in the study of sound change concerns the relationship between phonology and phonetics. Phonetic changes progress as continuous movement through the phonetic space, raising a number of questions about how a phonological representation is related to a phonetic realization if that phonetic realization can change. More importantly, however, is the question about when phonological representation is affected by, or affects phonetic change. The hypothesis of the Neogrammarians was that phonetic change was driven, at least at the outset, by pure phonetic conditioning, an idea which is fundamental to many reasearchers' models of sound change, including Ohala's (1981) model of hypo/hyper-correction, Janda & Joseph's (2003) Big Bang model, Blevins' (2004) Evolutionary Phonology, Bermudez-Otero's (2007) Lifecycle of Phonological Change, and many others'. In addition, the relatively recent influence of Exemplar Theory on the field of sociophonetics has deemphasized the role of abstract phonology in cognition as a whole, and in sound change in particular. On the other hand, there has also been a long tradition of treating phonological pressures as central to patterns of phonetic change. Martinet (1952) argues that the tension between a (perhaps teleological) pressure for phonological symmetry and the fundamental assymetry of the articulatory system drives a constant cycle of sound change.

I propose to direct my dissertation at some of these issues outlined above. I will argue that abstract phonological representations have an early, and continuing effect on phonetic change. I will be drawing most heavilly from the Philadelphia Neighborhood Corpus of Ling560 Studies, developed here at Penn. The interview catalogue contains recorded speech collected between 1973 and 2010, with 327 Philadelphians, whose dates of birth range from 1888 to 1991. The FAVE suite for forced alignment and automated vowel analysis has produced a corpus of 792,920 vowel measurements. It is on the basis of these measurements that my preliminary results are based.

To demonstrate the early effects of phonological representations on phonetic change, I focus on conditioned sound changes, specifically the raising of /ay/ before voiceless segments. /ay/-raising has been argued to have phonetically natural precursors (Joos, 1942; Chambers, 1973; Moreton & Thomas, 2004), but currently applies opaquely in the phonetically unnatural context of a flap which is underlyingly voiceless. Depending on how one wants to analyze the phonetic effects of a following flap, we should expect to observe a stage of /ay/-raising in Philadelphia where flaps pattern together, regardless of underlying voicing. However, it appears as if flaps have always patterned according to their underlying, phonological voicing, rather than their surface phonetic properties. This result has a number of consequences for theories of sound change. First, it means that the phonological innovation, the addition of a phonological process which creates two allophones of /ay/, occurred simultaneously with the beginning of the phonetic change. This is a fundamentaly different sequencing of events from the consensus view of phonetic change, where a period of pure phonetic conditioning precedes the phonological reanalysis. Secondly, it raises questions about the possibility of phonological rule insertion.

To support the continuing effects of phonology, I will present some preliminary results indicating diachronic cue trading within the allophones of /ay/ in Philadelphia, and some analysis of parallel shifts. It appears that as the pre-voiceless /ay/ allphone raises vowel height, the pre-voiced allphone is getting shorter. This would appear to be a trading of cues to their contrast, which is a fundamentally phonological contigency. Parallel phonetic changes, where members of a phonological natural class move in parallel, but do not fill a phonetic gap left behind by other members, indicate that the very objects of phonetic change are phonological features.

Proposal Manuscript
Slides (view in Adobe Reader for animations)
Slides source (knitr Rnw)
R source code