Examples of the spatial perturbations to the auditory feedback of the F2 trajectory. A, A schematic diagram showing the setup for auditory feedback perturbation. B, An example spectrogram of the stimulus utterance “I owe you a yo-yo” with the F1 and F2 trajectories (dashed black lines) overlaid and the F2 extrema labeled. The blue vertical lines highlight the focus interval, that is, the interval during which the auditory perturbation occurs. C, A zoomed-in view of the focus interval of the example in B. D, E, Resultant spectrograms of Down (D) and Up (E) perturbation on the same sound as shown in C. The yellow dashed lines show the shifted F2. For comparison, the black dashed lines in D and E are identical to those in C. F, G, Examples of the temporal perturbations Accel and Decel, in the same format as D and E.
Schematic illustrations of the spatial (Down and Up) and temporal (Accel and Decel) perturbations. A, The mapping between the original and perturbed values of F2 in the spatial (Down and Up) perturbations. B, The time-warping functions used in the temporal (Accel and Decel) perturbations.
Articulatory compensations under the spatial (Down and Up) perturbations. A, Grand average F2 trajectories aligned at the F2 maximum in [i] of “I”. The time axis shows unnormalized (real) time and includes only an early part of the utterance, from [i] to [u]2. Error bands are omitted for clarity of visualization. F2 was amplitude normalized before averaging across subjects. B, Average F2 trajectory changes from the noPert baseline under the Down and Up perturbations. The thin lines show mean ± 1 SEM. The time axis is the same as that of A. The magenta bars indicate the intervals of significant difference between the responses to Down and Up perturbation under paired t tests with a statistical threshold of FDR = 0.05. The blue bar shows the comparison between the Down and noPert conditions (FDR = 0.05). C, Average F2 magnitude changes shown on the segment-normalized time axis (for details on the time normalization procedure, see Results). The meaning of the magenta and blue bars are the same as in B; the red bar indicates the comparison between the Up and noPert conditions (FDR = 0.05). The arrows signify the correspondence with the data shown in D–G, which show spatial changes under the perturbations. D, Change in the value of F2 at the minimum in [u]1. E, Change in value of F2 at the temporal midpoint between the F2 minimum in [u]1 and the F2 maximum in [j]1. F, Change at the F2 maximum in [j]1. G, Change at the midpoint between the F2 maximum in [j]1 and the F2 minimum in [u]2 in “you”. Error bars indicate ±1 SEM. H, I, Timing changes under the perturbations. H, Change in the [i]–[u]1 time interval. I, Change in the [i]–[j]1 interval. Asterisks, Significant difference at p < 0.05 (post hoc Tukey's HSD following RM-ANOVA). FB, Feedback.
Articulatory adjustments under the temporal (Accel and Decel) perturbations. A, Grand average (across trials and subjects) of F2 trajectories aligned at the F2 maximum at [i]. The format is the same as Figure 3A. The solid curves show production; the dashed curves show auditory feedback. The magnitude of the F2 at the [u]1 minimum under the Decel perturbation (dashed green curve) is apparently altered from the value in the production because the timing of the [u]1 minimum varies across different trials and different subjects. In individual trials, the F2 magnitudes at this minimum were always preserved by the Decel perturbation (see B). B, C, Articulatory timing changes under the perturbations. B, Change in the [i]–[u]1 interval (error bars are +1 SEM). C, Change in the interval between the [i]–[j]1. Asterisks, Significant difference at p < 0.05 (post hoc Tukey's HSD following RM-ANOVA). FB, Feedback.
Changes in articulatory timing beyond the vicinity of the focus interval. Changes in the timing of the six major F2 landmarks ([u]1, [j]1, [u]2, [j]2, [u]3, and [j]3; see Table 1) under the Accel and Decel perturbations. The filled symbols represent significant difference from the baseline (t test, p < 0.025); the asterisks indicate significant difference between the Accel and Decel conditions (paired t test, p < 0.025).
Ad hoc phonetic symbols used in the current paper to denote the F2 extrema in the utterance “I owe you a yo-yo” and the baseline values of the F2 at these extrema and their timing with respect to the first maximum ([i])