Modeling Phoneme and Open-Set Word Recognition by Cochlear Implant Users Based on Psychophysical Performance: A Preliminary Report1

We have recently made some progress in our ability to explain perceptual performance by adult cochlear implant (CI) users, and this research may eventually be applicable to children. Quantitative, psychophysically-based models of vowel (Svirsky & Meyer, 1997) and consonant (Meyer, Svirsky, Castor, Falsafi & Simmons, 1997) perception generated predicted confusion matrices that were similar to matrices averaged across seven adult CI users (Skinner, Holden & Holden, 1995). Frisch and colleagues (Frisch & Pisoni, 1997; Frisch, Meyer, Pisoni, Svirsky & Kirk, submitted) have developed models to predict open-set word recognition by CI users based on their vowel and consonant feature recognition. Using the Frisch et al. model, we obtained predictions of open-set spoken word recognition from observed vowel and consonant confusion matrices for a group of adult CI users, and we compared these predictions to observed performance on a test of open-set spoken word recognition. In the future, we will use both models to predict open-set spoken word recognition from performance on psychophysical tests for individual CI users, including children.