Schematic F0 contours for the three sets of manipulations performed on the original breathy-toned stimuli. The upward facing arrow indicates the direction of F0.
Schematic F0 contours for the two sets of manipulations performed on (a) the original low modal-toned stimuli, and (b) the original creaky-toned stimuli. The upward-facing arrow indicates increasing F0, and the jagged lines represent creaky phonation.
(Color online) (a) Proportion “breathy-toned” responses for breathy stimuli as a function of start of F0 fall. (b) Proportion “creaky-toned” responses as a function of start of F0 fall, for short original low modal (22) and low-falling creaky (21) tokens. Chance is indicated at 0.14. Best-fit regression lines are shown for (a) proportion breathy response as a function of F0 fall (excluding subject-explained variance), and (b) for shortened modal (grey line) and short creaky (black line) stimuli. The starting value of the F0 fall is weakly correlated with proportion “breathy-toned” responses. For “creaky-toned” responses, F0 fall shows moderate to high correlations with proportion “creaky-toned” responses for both modal and creaky stimuli, as indicated by Pearson's r values. Note that these correlations are not derived from the logistic mixed-effects models.
(Color online) Schematics of the spectral source manipulations, with only the five harmonic anchors (H1, H2, H4, 2 kHz, and 5 kHz) represented by vertical lines. The four component slopes of the source spectrum model are shown by dotted lines. In condition 1 (a), H1–H2 varied by increasing the amplitude of H1. In condition 2 (b), H1–H2 and H2–H4 covaried by decreasing the amplitude of H2. In condition 3 (c), H2–H4 and H4–2 kHz covaried by decreasing the amplitude of H4. Manipulations are represented by dashed lines.
Proportion of “breathy-toned” responses as a function of (a) H1–H2 in conditions 1 (H1–H2 varying) and 2 (H1–H2 and H2–H4 covarying), (b) H2–H4 in conditions 2 and 3 (H1–H2 held constant, H2–H4 and H4–2 kHz covarying). Best-fit regression lines for proportion of “breathy-toned” responses as a function of changes to (a) H1–H2 and (b) H2–H4 are included, excluding subject-explained variance. H1–H2 manipulations in condition 1 but not in condition 2 are highly correlated with a higher proportion of “breathy-toned” responses (as indicated by the values of Pearson's r). H2–H4 manipulations in condition 3 but not in condition 2 are highly correlated with a higher proportion of “breathy-toned” responses. Note that these correlations are not derived from the logistic mixed-effects models.
Overview of White Hmong tones, from Esposito (2012) .
Mean values of H1*–H2*, H1*–A1* (asterisks indicate that the measures are corrected for formant frequencies and bandwidths), and HNR in dB (standard deviations in parentheses) for high-falling breathy vs modal and low creaky vs low modal stimuli, across all pitch manipulations.
Fixed-effects results of logistic model predicting “breathy-toned” responses.
Fixed-effects results of logistic model predicting “creaky-toned” responses.
Summary of the five conditions in experiment 2. Arrows indicate an increase (↑) or decrease (↓) for a given component. An unmentioned component means it was held constant in that condition.
Fixed-effects results of logistic model predicting “breathy-toned” responses in experiment 2.
Fixed-effects results of logistic model predicting “/ɔ/ word” responses in experiment 2.
Article metrics loading...
Full text loading...