Table 1.
Learning performance for successful runs
| Language | No. of speakers w/successful runs* | Average no. of successful runs* | Median percent correct† | Median d′ for length discrim.† | Median d′ for spectrum discrim.† |
|---|---|---|---|---|---|
| Parametric model, OME | |||||
| English | 18 of 19 | 7.7 ± 2.8 | 92.7 (93.4) | 3.91 (3.90) | 3.19 (3.22) |
| Japanese | 10 of 10 | 7.9 ± 3.0 | 91.1 (91.9) | 4.09 (4.09) | 3.32 (3.30) |
| Nonparametric model, TOME | |||||
| English | 18 of 19 | 5.4 ± 2.9 | 83.0 (91.3) | 3.78 (3.83) | 2.70 (3.06) |
| Japanese | 10 of 10 | 5.5 ± 1.6 | 85.2 (91.2) | 4.05 (3.98) | 3.11 (3.25) |
*Speakers with successful runs, with 10 runs per speaker.
†Percent-correct and d′ values are medians across speakers of the average over successful runs within a speaker. Parenthetical values show supervised training results.