Mechanisms of Phonological Change

The traditional Philadelphia allophonic /æ/ system (henceforth: PHL shown in (1) below) is characterized by a set of complicated conditioning factors and a dramatic acoustic distinction between the two allophones. In recent years, some Philadelphians have begun to exhibit a new allophonic system (NAS, shown in (2) below). Like PHL, NAS is characterized by a dramatic acoustic distinction between tense and lax allophones. NAS is quickly overtaking PHL in the Philadelphia community, as demonstrated by Labov et al. (2016). (1) PHL: æ → æh/ _ [+anterior] ∩ ( [+nasal] ∪ [-voice + fricative) ]σ (2) NAS: æ → æh/[+nasal ] This situation offers an exciting opportunity to observe phonological change in individual speakers. Most phonological changes involve the collapse or creation of a new phonological category; because of the large degree of acoustic overlap in these situations, it is difficult or impossible to identity individual tokens as having been produced by the old or the new phonology. In the current change in Philadelphia /æ/, however, both the old and the new system involve distinct acoustic targets, making it possible to identify which underlying system was used to produce a given word. It is therefore possible to test several distinct theories about phonological change: Whether change occurs through gradual phonetic incrementation (e.g. Ohala 1981), through individual speakers producing only the old or the new system (e.g., Janda and Joseph 2003), or whether change occurs via individual speakers probabilistically producing both the old and the new system in a process of individual grammar competition (e.g., Fruehwald et al. 2013). In my dissertation, I examine natural speech production from 46 speakers who acquired language during the period of allophonic change, with a combination of topic-directed conversations and targeted natural language experiments. Using a glm classifier, I identify tokens of /æ/ as having been produced by either PHL or NAS. In concert with an analysis of speakers’ social histories, I use these results to argue that the change from PHL to NAS in Philadelphia is driven by the mechanism of competing grammars, suggesting that both syntactic change and phonological change proceed in the same manner. My research provides one of the first pieces of direct empirical support for a unified theory of language change in which structural changes in syntax and phonology are implemented through the same mechanism of grammar competition (Kroch, 1989; Fruehwald et al., 2013). In addition to the theoretical contribution to phonological change, my dissertation also traces the social patterns of the allophonic change, highlighting the effect of network structure and access to elite education on the adoption of the incoming allophonic system. I also employ experimental methods to demonstrate that the abstract allophonic rules of /æ/ are the target of social evaluation and contribute to social meaning. I find speakers producing surprisingly systematic evaluations of PHL and NAS, a result which only emerges when analyzing the evaluation of changing abstract parameters. Finally, to test whether the change from PHL to NAS was the inevitable result of phono- logical simplification, I developed a computational simulation built using a principle of language acquisition (Yang, 2016) to demonstrate that the allophonic restructuring in /æ/ was not the result of children simplifying their input data, but rather must have been the result of dialect contact with in-moving speakers of the new system