The stimuli for the vowel identification experiment were sampled from a two-dimensional spectral grid defined by the center frequencies of the two lowest resonances in the human vocal tract, F1 (representing the vowel height dimension) and F2 (as a correlate of backness and lip rounding). F1 values ran from 2.5 to 8.5 Bark, in 7 steps of 1 Bark. The F2 values were varied in 9 steps of 1 Bark between 6 and 14 Bark. This yields 9 × 7 = 63 different vowel spectra in a rectangular matrix, as is illustrated in Figure 5. The use of equal steps (of 1 Bark) ensures that each step yields a change of vowel quality that is perceptually the same (Traunmüller, 1991). A number of F1-by-F2 combinations cannot be produced by the human vocal organs, and when synthesized, sound inhuman. These impossible/inhuman F1-by-F2 combinations were eliminated from the set of 63, so that a subset of 43 legitimate combinations remained, which together constitute a uniformly sampled universal acoustic vowel triangle. For details of the stimulus synthesis and motivation of choices made, we refer to Van Heuven et al. (2020).