Electromagnetic recordings reveal latency differences in speech and tone processing in humans.

Human auditory electromagnetic brain responses to sinusoidal (tone) and speech stimuli (Finnish vowel /a/ with two different glottal excitations) were studied with whole-head magnetoencephalogram (MEG) and electrodes placed on the subject's scalp. The frequency and intensity of the sinusoidal stimulus were optimally adjusted to match the spectra of the speech stimuli. Both tone and speech sounds elicited a prominent electric N1-P2 and magnetic N1m-P2m response complex. N1 and P2 amplitudes were larger to speech sounds than those to the tone. This amplitude enhancement was not as evident in the N1m and P2m obtained in MEG. Both the N1(m) and P2(m) latency always peaked earlier for the tone than for the vowels. The source origin of N1m for both the tone and speech stimuli was in the auditory cortex, there being no significant location differences as a function of stimulus type. N1m in the right hemisphere was anterior to that on the left, and P2m was anterior to N1m in both hemispheres. Varying the perceptual quality of the vowels by changing their glottal excitations (from "soft" /a/ to "pressed" /A/) had no effect on the response amplitudes or latencies. Thus, the present results show that only the latency behavior of N1(m) and P2(m) reliably dissociates speech and tone processing in humans. The findings are discussed in relation to previous observations on cortical processing of sinusoidal and vowel sounds and with regard to the glottal excitation in speech processing.