This dissertation addresses the broad question about how phonology and phonetics are interrelated, specifically how phonetic language changes, which gradually alter the phonetics of speech sounds, affect the phonological system of the language, and vice versa. Some questions I address are:
To address these questions, I drew data from the Philadelphia Neighborhood Corpus [PNC] (Labov and Rosenfelder, 2011), a collection of sociolinguistic interviews carried out between 1973 and 2013. Using the PNC data, I utilized a number of different statistical modeling techniques to evaluate models of phonetic change and phonologization, including standard mixed effects regression modeling in R (Bates, 2006), and hierarchical Bayesian modeling via Hamiltonian Monte Carlo in Stan (Stan Development Team, 2012).
My results are challenging to the conventional wisdom that phonologization is a late-stage reanalysis of phonetic coarticulatory and perceptual effects (e.g. Ohala, 1981). Rather, it appears that phonologization occurs simultaneously with the onset of phonetic changes. I arrive at this conclusion by examining the rate of change of contextual vowel variants, and by investigating mismatches between which variants are expected to change on phonetic grounds versus phono- logical grounds. In my analysis, not only can a modular feed-forward model of phonology and phonetics account for observed patterns of phonetic change, but must be appealed to in some cases.
These results revise some the facts to be explained by diachronic phonology, and I suggest the question to be pursued ought to be how phonological innovations happen when there are relatively small phonetic precursors.
In this chapter, I first disscuss the relationship between sound change and grammar, specifically how your theory of speakers' knowledge of their sound systems defines and circumscribes possible sound changes. I then lay out my minimal theoretical committments, and discuss my proposal that phonetic change ought to be understood as changing phonetic implementation. I compare quantitative predition of my proposal against a competing proposal that all sound change is phonological.
This is my favorite figure from this chapter, which outlines the quantitative prediction of what phonetic change would look like if it was actually phonological.
In this chapter, I describe the data that I use throughout the dissertation. It consists of vowel formant measurements drawn from the Philadelphia Neighborhood Corpus.
My favorite figure from this chapter is a histogram of how many measurements we have per-speaker. It shows that most speakers have just a bit more than 1,000 vowel measurements.
Telling the difference between phonetic and phonological processes is notoriously difficult. In this dissertation, I attempt to distinguish between the two by examining the rate of change of different vowel variants. I employ a Bayesian Heirarchical Model to do so. According to this method, during phonetic change, phonetic variants rarely become phonologized. In fact phonologization appears to happen at the onset of these phonetic changes in every case.
My favorite figure from this chapter displays the predicted trajectories of /ow/ variants along F2. Note how [owL], representing /ow/ before /l/, is markedly divergent from all the other /ow/ variants from the very beginning.
First, I look at a few cases where there is a mismatch between the predicted conditioning of phonetic change on phonetic grounds versus phonological grounds. In the examples available from the Philadelphia neighborhood corpus, it appears as if phonological conditiong is at work from the very onset of the changes, rather than being late-stage reanalyes.
Then, I look at parallel phonetic shifts, which involve the wholesale, simultaneous phonetic change of a phonological natural class.
It's a tie for my favorite figure from this chapter. First is the result of a model which shows that pre-voiceless /ay/ raising (a.k.a Canadian Raising) has always applied before flapped /t/ and /d/ in the same way as it does before unneutralized /t/ and /d/.
Second is this plot showing the diachronic trajectory of just 'day' and 'days', showing that /ey/ raising in Philadelphia is an active phonological process.
In this chapter, I argue against a model of gradual phonologization wherby small phonetic errors gradually accumulate into a new phonological process or category. Based on the results from Chapters 4 and 5, it really looks like new phonological processes enter the grammar at the onset of phonetic changes. Of course, this is somewhat controversial, and the suggestion that new phonological processes enter the grammar before they have robust phonetic correlates may strain credulity. So, I spend most of the rest of the chapter further arguing for the plausibility of this scenario.
My favorite figure from this chapter shows the conditioning on /ey/ raising that I discovered in Chapter 5. Surprisingly, even though a following /l/ appears to most favor the change at the beginning, /ey/ actually never undergoes the change in this context.
This is just a very brief conclusion to the dissertation, and it contains no figures from which to choose a favorite.