Effects of talker uncertainty I: Auditory word recognition

The production and resulting acoustic composition of spoken words vary as functions of individual talker characteristics. However, the effects of talker differences on auditory word recognition processes have been largely ignored by researchers working in speech perception. In the present study, the effects of talker differences on perception of spoken words were examined by manipulating two factors: talker uncertainty (i.e., words produced from a single talker or from 15 talkers) and lexical structure (i.e., high- or low-density lexical items). In the first experiment, identification performance for words presented in noise was worse when the items were produced by different talkers, than by a single talker. In the second experiment, latencies and accuracy in a naming task were also worse under multiple-talker condition. In both experiments, lexical density alone did not have a significant effect on performance. The relationship of these results to the underlying processing operations involved in auditory word recognition will be discussed. [Work supported by NIH.]