Department of Linguistics
University of Pennsylvania
Changes in the Philadelphia accent from the perspective of natural selection
Next to being used as a system for communication, language additionally conveys a large amount of social information. In the literature on language change, there is agreement that this social function of language plays a key role in the transmission of language change (e.g. Labov 2001; see Nesbitt 2023 for an overview). At the same time, language is subject to internal pressures. For example, the vowels in FIGHT and MOUTH are naturally pronounced more raised (sounding a bit like “FOIGHT” or “MEUTH”) due to their following consonants, which have been argued to induce biomechanical bias in the articulation of those vowels through two possible mechanisms (Davis & Berkson 2021). What is the role of these different pressures in processes of language change? Is language change primarily driven by these social considerations, driven by these internal pressures, or is it driven by neither but do these merely limit the space for possible changes, and do possible changes drift through this space randomly (‘stochastic drift’)? I investigate this question for language change in the vowels of Philadelphia English, with a current focus on the aforementioned FIGHT/MOUTH-raising. I analyze the vowel data in the Philadelphia Neighborhood Corpus (408 speakers born in 1880-1994), using a statistical model from population biology that disambiguates between stochastic drift and directional selection (the ‘driving’ of change). Results show a strong effect of selection pressure, which moreover is significantly larger in women than in men (against Conn 2005). Most importantly, selection on the vowels is significantly stronger in the FIGHT/MOUTH context than in contexts with different following consonants, supporting the hypothesis that these exert some internal pressure. At the same time, however, I fail to find support for either of the proposed sources of this pressure proposed by Davis & Berkson (2021). I discuss the implications of this mixed evidence for the bigger picture of language change, and I sketch preliminary ideas for an experiment that might help further elicudate this picture.