Speech Specific Auditory Evoked Potentials

ERPs of 27 scalp electrodes were recorded from fourteen volunteers during the presentation of binaurally presented speech and non-speech stimuli. Speech stimuli consisted of nine different consonant-vowel syllables. Non-speech stimuli were generated using special envelope-modulated low pass noise. With this method, kinematic signals are time and amplitude normalized, allowing us to observe the processing of the underlying spectral cues. We presented the speech and non-speech stimuli in three different conditions to examine the influence of the presentation order (alternating vs. randomized) in two passive conditions and the influence of attention (active vs. passive) in one active condition. The ERPs to speech-stimuli were enhanced, in contrast to the ERPs to noise stimuli, independently of the type of stimulus presentation. The results give evidence for an early electrophysiological correlate of distinct neuronal processing of speech and noise stimuli. The present study shows that this type of noise-stimuli seems to be a useful tool for the comparison of the processing of speech. One aim of speech processing research is to extract acoustic factors which are important for the integration and segregation of speech stimuli in the central auditory pathway. For instance, understanding basic structure-function relationships may help to develope more efficient hearing aids and speech analysing systems. One method to measure auditory processing online are event-related brain potentials (ERPs). An advantage of using ERPs is the possibility to measure responses of the information processing system even in situations when subjects do not attend to the stimuli. Therefore automatic and pre-attentive processes can be separated from the attentive processing of information (Näätänen, 1992). Differences in the processing of variations in frequency, intensity and duration can be examined with this method by using simple acoustic stimuli (Schröger, 1998). Speech stimuli are more complex than other auditory stimuli, such as pure tones. In continuous speech we can find a highly spectral order in the observation of frequency, and intensity movement over time. This is problematic because the complexity of a stimulus is known to affect the ERPs (Tervaniemi et al., 2000) and holds especially for slight spectral changes in formant transitions (e.g. Kraus et al., 1995, Maiste et al., 1995). The aim of the present study was to find a useful method to observe the information of continuous speech which implies a pronounced temporal structure (Zwicker & Fastl, 1998). One possibility of extracting temporal parameters of speech is to preserve the intensity contour of the speech signal. In recent studies using this kind of control stimuli, a speechspecific component at 170 ms (N 170) after stimulus onset was observed (Rosanowski et al., 1996, Hies, 2001). This component occurs for healthy subjects, but not in patients with a special disturbance of speech perception, such as Global and Wernicke’s aphasia (Rosanowski et al., 1996). The purpose of this study was to replicate the speech specific cortical potentials after acoustic stimulation with different consonant vowel stimuli using the method of Hies (2001), i.e. loudness controlled speech and noise stimuli. In order to test the generality of the speech specific ERP component we compared different types of stimulus presentation order and the influence of attention on the processing of this kind of stimuli.