| Popis: |
Previous studies have employed signal detection theory to analyze data from speech and nonspeech experiments. Typically, signal distributions were assumed to be Gaussian. Schouten and van Hessen [J. Acoust. Soc. Am. 104, 2980–2990 (1998)] explicitly tested this assumption for an intensity continuum and a speech continuum. They measured response distributions directly and, assuming an interval scale, concluded that the Gaussian assumption held for both continua. However, Pastore and Macmillan [J. Acoust. Soc. Am. 111, 2432 (2002)] applied ROC analysis to Schouten and van Hessen’s data, assuming only an ordinal scale. Their ROC curves suppported the Gaussian assumption for the nonspeech signals only. Previously, Lopez‐Bascuas [Proc. Audit. Bas. Speech Percept., 158–161 (1997)] found evidence with a rating scale procedure that the Gaussian model was inadequate for a voice‐onset time continuum but not for a noise‐buzz continuum. Both continua contained ten stimuli with asynchronies ranging from −35 ms to +55 ... |
