Acoustic differentiation of allophones of /aɪ/ in Chicagoland English: Statistical comparison of formant trajectories

Diphthongs have a dynamic formant structure. Nevertheless, many quantitative studies of diphthongs are based on measurements at only two points, somewhere in the nucleus and somewhere in the glide. The question arises as to whether analyses based on values at only two points provide an adequate understanding of the dynamics of diphthongs. Wieling (2018) mentions the analysis of /aɪ/ raising in Chicago English in Hualde, Luchkina & Eager (2017) as one of several examples of recent studies where potentially interesting patterns may have been missed because of limited sampling of formant values, and proposes using Generalized Additive Mixed Models (GAMM) to allow a more complete understanding of diphthong dynamics. In this paper, we examine the acoustic nature of the (quasi-)phonemic differentiation between two originally allophonic variants of the diphthong /aɪ/ in the US English of Chicago and the surrounding area. We offer an acoustic analysis based on full formant trajectories of diphthongs with data obtained from a group of 53 speakers. The results of a GAMM analysis are then compared with those obtained in Hualde et al. (2017), which considered values at only two points and from a smaller set of speakers (17). We also discuss the main advantages of GAMM analysis over other techniques that have being proposed for the analysis of differences in vowel formant dynamics.