Fig 1.
Example displays (no-contrast, 1-contrast vs. 2-contrast conditions) for the instruction, “Click on the large cup”.
The no-contrast condition was used in Grodner and Sedivy [18], but replaced with 2-contrast condition in the current experiment. This figure is original, compliant with the CC BY4.0 license, produced by the current study team based on the information provided by Grodner and Sedivy [18]. The images for the 1-contrast and 2-contrast visual displays are also original. They are similar, but not identical, to those that were used in the eye-tracking studies. They are presented here for illustrative purposes only.
Fig 2.
A. Time-course of fixations during instructions relative to the onset of the adjective (e.g., Click on the large cup) in the reliable (top) and unreliable (bottom) speaker conditions in Experiment 1. Vertical lines indicate the average adjective onset, adjective offset, and noun offset. Shaded area indicates the 500 ms planned analysis window. Points represent averages of by-participant means in each time bin. Error bars represent 95% bootstrapped confidence intervals over those by-participant means. B. Average duration of the scalar adjectives and the final nouns and their alignment with the 500ms planned analysis window.
Table 1.
Summary of final linear mixed model of fixation proportions in Experiment 1.
Table 2.
Combinations of instructions and bottom-up manipulations in Experiments 1 and 2.
Fig 3.
Time course of target fixations during instructions (e.g., Click on the large cup) in the unreliable-speaker condition with only bottom-up information (Experiment 2).
The 1- and 2-contrast conditions are plotted in the left and right panels, respectively. Vertical lines indicate the mean 1) adjective onset (aligned to 0), 2) adjective offset, and 3) noun offset. Shaded area indicates the 500ms planned analysis window.
Fig 4.
Average proportions of target fixations during the 500ms window starting 200ms after the adjective onset (e.g., Click on the large cup) by reliability and contrast conditions in Experiments 1 and 2.
Error bars represent 95% bootstrapped confidence intervals over those by-participant means.
Table 3.
Helmert coding of reliability conditions.
Table 4.
Summary of final linear mixed model of fixation proportions in Experiments 1 and 2.
Reliability conditions were Helmert coded (for details, see text).
Fig 5.
Average proportions of target fixations during the 500 ms window starting 200 ms after the adjective onset (e.g., “Click on the large cup”) by reliability and contrast conditions in Experiment 1.
Error bars represent 95% bootstrapped confidence intervals over those by-participant means.