Back Vowel Dynamics and Distinctions in Southern American English

Abstract

Southern American English is spoken in a large geographic region in the United States. Its characteristics include back-vowel fronting (e.g., in goose, foot, and goat), which has been ongoing since the mid-nineteenth century; meanwhile, the low back vowels (in lot and thought) have recently merged in some areas. We investigate these five vowels in the Digital Archive of Southern Speech, a legacy corpus of linguistic interviews with sixty-four speakers born 1886-1956. We extracted 89,367 vowel tokens and used generalized additive mixed-effects models to test for socially-driven changes to both their relative phonetic placements and the shapes of their formant trajectories. Our results reinforce previous descriptions of Southern vowels while contributing additional phonetic detail about their trajectories. Goose-fronting is a change in progress, with greatest fronting after coronal consonants. Goat is quite dynamic; it lowers and fronts in apparent time. Generally, women have more fronted realizations than men. Foot is largely monophthongal, and stable across time. Lot and thought are distinct and unmerged, occupying different regions of the vowel space. While their relative positions change across generations, all five vowels show a remarkable consistency in formant trajectory shapes across time. This study’s results reveal social and phonetic details about the back vowels of Southerners born in the late nineteenth and early twentieth centuries: goose-fronting was well underway, goat-fronting was beginning, but foot remained backed, and the low back vowels were unmerged.

Perturbing the consistency of auditory feedback in speech

“…However, talkers may be using other aspects of vowels to control articulation than static indices of formant frequency. Vowels have inherent formant dynamics that vary with dialect, age, and gender of speakers (e.g., Stanley et al, 2021 ). These dynamics can influence compensatory behavior with participants correcting for changes in spectral trajectories ( Jibson, 2020 ).…”

Boomer Peak or Gen X Cliff? From SVS to LBMS in Georgia English

“…For speakers from LAGS/DASS, transcription was manual and checked for errors up to four times (see pipeline in Olsen et al, 2017). Forced alignment was done using the Montreal Forced Aligner ( MFA ; McAuliffe, Socolof, Mihuc, Wagner, & Sonderegger, 2017), which is estimated to have a word-level accuracy rate of 86% for this data (Stanley et al, 2021:398). Formant values and other acoustic measurements were extracted at five time points (20%, 35%, 50%, 65%, 80% of vowel duration) using a local installation of the FAVE suite (Rosenfelder, Fruehwald, Evanini, Seyfarth, Gorman, Prichard, & Yuan, 2014).…”

“…We evaluate the GAMMs in two ways: first, via visualizations of predicted measurements, which were extracted from each model. Second, we tested the significance of YoB via model comparison (Renwick & Stanley, 2020; Stanley et al, 2021). For each GAMM, we constructed a model that excluded YoB but was otherwise identical.…”

Australian English Monophthong Change across 50 Years: Static versus Dynamic Measures

“…Such dynamicity holds the key to phonemic identity in the case of diphthongs, but research shows that vowel inherent spectral change (VISC) is a feature of vowels more generally (e.g., Nearey and Assmann 1986;Nearey 2013). Many studies have shown that VISC has a sociophonetic function (see e.g., Fox 2011, 2013;Docherty et al 2018;Farrington et al 2018;Kirkham et al 2019;Sóskuthy et al 2019;Renwick and Stanley 2020;Stanley et al 2021). Dynamic characteristics of vowels also change diachronically (see e.g., Jacewicz and Fox 2013;Gubian et al 2019;Harrington et al 2019a;Sóskuthy et al 2019;Cox et al, forthcoming), and such patterns of change may be obscured by purely static analyses.…”