Perception of the American English liquid /ra-la/ contrast by humans and monkeys

Human and monkey perception of the American English liquid /ra-la/ contrast was compared using various synthetic continua in which the normal spectral and temporal cues were higher either complete, partial, or altered in various ways. Two experiments compared human and monkey discrimination of the various continua using a low-uncertainty repeating-standard procedure. Results showed that, while human sensitivity was best at the human phoneme boundary, monkey sensitivity was best inside the /ra/ category. Also, while humans were more sensitive than monkeys to temporal variation in the stimuli, monkeys were more sensitive than humans to spectral variation, particularly for stimuli inside the /ra/ category. Two additional experiments compared human and monkey identification of the /ra-la/ continua using a higher-uncertainty go/nogo identification procedure. Monkeys performed as accurately as humans (including one native Spanish and one native Hindi listener) in identifying the stimuli. However, human and monkey "phoneme boundaries" were in different places, with monkey boundaries shifted more toward /ra/ than human boundaries. These results suggest that human boundaries may be based on some sort of specific linguistic knowledge. Despite these boundary differences, monkeys showed a trading relation comparable to that of humans, indicating that the /ra-la/ trading relation has a psychoacoustic basis and that linguistic knowledge is not a necessary prerequisite for it to occur.