Relationship between the discrimination of (w-r) and (t-d) continua and the identification of distorted (r)

The purpose of this study was to assess the extent to which listeners can perceive intraphonemic differences. In Experiment 1, subjects identified synthesized acoustic tokens of child-like speech that varied in second and third formant (F2 and F3) onset frequencies as (w), (r), or distorted (r) in two conditions: (a) with and without feedback of the group response choices, and (b) before and after training to identify the best examples of (w), (r), and distorted (r) based on their identification in the first condition. The results were: (a) some subjects consistently identified distorted (r) above criterion, and (b) feedback was more effective in increasing distorted (r) identification than was training. In Experiment 2, the same subjects participated in discrimination tasks using stimuli from a synthesized child (w-r) continuum that varied in F2 and F3 onsets and from a synthesized adult (t-d) continuum that varied in preconsonantal vowel duration. The results were: (a) perception was not categorical for both continua, (b) little relation was found between distorted-(r) identification and measures of (w-r) discrimination, and (c) a high and significant correlation was found between identification of distorted (r) and within-(d) discrimination. In Experiment 3, different subjects identified the child manifold stimuli and discriminated stimuli in a synthesized child (w-r) continuum and in a synthesized adult (t-d) continuum. The results were: (a) neither (w-r) or (t-d) perception was categorical although the former came closer than the latter in terms of individual subject performance, (b) there was a high and significant correlation between distorted-(r) identification and within-(r) discrimination of (w-r) stimuli, and (c) there were high and significant correlations between distorted-(r) identification and mean, cross-category boundary, and within-(t) discrimination of (t-d) stimuli.