FIGURE 4.
The unimodal (top) and bimodal (middle) training distributions of F1 values used in the present experiment, with an equal range and a nearly equal SD and edge strength (explanation: see text). The unimodal distribution represents the Spanish vowel /a/ and the bimodal distribution is representative of the Dutch vowel contrast /ɑ/∼/a/. Each vertical line shows the F1 value of a single stimulus. (For the purpose of clarity only 64 values are shown, rather than the 256 values used). Test stimuli in the present experiment (bottom). The F1 values of the test stimuli lie at the intersections of the two distributions.