The medium-term dynamics of accents on reality television: Supplementary Material

Language 93.3, September 2017 s1 THE MEDIUM-TERM DYNAMICS OF ACCENTS ON REALITY TELEVISION: ONLINE SUPPLEMENTAL MATERIALS MORGAN SONDEREGGER MAX BANE PETER GRAFF McGill University and Centre for Research on Brain, Language, and Music University of Chicago University of Vienna 1. DYNAMIC MODEL RESULTS. Tables S1–S3 supplement the results of the dynamic models reported in§5 of the main text. Table S1 summarizes predicted time dependence in VOT for voiced and voiceless stops for each speaker: predicted average VOT (a speaker’s ‘offset’ in the static model), whether the model includes by-day variability and/or a time trend, and σ, which quantifies the degree of by-day variability. Table S2 summarizes the predicted time dependence in the CSD rate for each speaker (analogously to Table S1): predicted average CSD rate (the intercept of a speaker’s dynamic model), whether there is by-day variability and/or a time trend, and σ, which quantifies the degree of by-day variability (when present). Table S3 summarizes predicted time dependence in F1 and F2 for each vowel, for each speaker, similarly to VOT. VOICED STOPS VOICELESS STOPS SPEAKER MEAN BDV? σ TT? MEAN BDV? σ TT? Dale 16.79 T 0.17 F 66.96 T 0.09 T Darnell 10.81 T 0.22 F 47.06 T 0.10 F Lisa 20.79 T 0.09 F 55.84 T 0.03 F Luke 10.66 T 0.13 F 38.07 T 0.04 F Michael 16.50 T 0.17 T 62.58 T 0.08 T Mohamed 12.27 T 0.26 F 38.47 T 0.05 T Rachel 17.42 T 0.13 T 57.56 T 0.11 T Rebecca 15.88 T 0.17 T 50.78 T 0.06 T Rex 20.23 T 0.09 T 51.63 T 0.09 T Sara 18.07 T 0.21 F 49.14 T 0.16 F Stuart 16.17 T 0.12 F 68.52 T 0.11 F TABLE S1. Summary of predicted time dependence in VOT for each speaker for voiced and voiceless stops. Mean: model-predicted average VOT. T/F values are Boolean values describing which kind of time dependence a speaker shows (Fig. 1): none, by-day variability (BDV), time trends (TT), or both. σ measures the size of by-day variability. (See main text.) s2 TABLE S2. Summary of predicted time dependence in coronal stop deletion rate for each speaker. T/F columns are defined as in Table S1. σ measures the size of by-day variability. (See main text.) 2. PLASTICITY. Tables S4 and S5 give supplemental information for the phonetic plasticity analysis (§8). Table S4 shows the plasticity value for each variable/speaker pair. Table S5 shows Kendall’s τ, a measure of rank correlation, between the plasticities for each pair of variables (across speakers), with an associated p-value. (We use rank correlation because plasticity for different variables is on different scales.) With one exception (VOT for voiced and voiceless stops), all pairs of variables are positively associated (τ ∈ [0.05, 0.67]). However, these correlations are usually not significant (p > 0.05; exceptions: STRUT/TRAP, GOOSE/CSD). Thus, while there is a consistent trend for a speaker who varies more in a given variable to vary more in another, the trend for individual pairs of variables is usually weak. SPEAKER MEAN CSD RATE BDV? σ TT? Dale 0.62 T 0.15 F Darnell 0.68 T 0.22 F Lisa 0.52 F — F Luke 0.43 T 0.08 F Michael 0.51 T 0.12 T Mohamed 0.67 T 0.20 F Rachel 0.59 T 0.05 F Rebecca 0.48 F — T Rex 0.52 T 0.15 F Sara 0.60 F — T Stuart 0.61 T 0.13 T s3 F1 + F2 F1 F2 VOWEL SPEAKER BDV? TT? MEAN BDV σ TT MEAN BDV σ TT GOOSE Dale T T −1.02 T 0.00 T 0.40 T 0.08 F Darnell T T −1.25 T 0.14 F 0.05 T 0.14 T Kathreya T F −0.86 T 0.17 F −1.23 T 0.09 F Lisa T...