Psycholinguistic models of recognition of speech sounds