Article révisé par les pairs
Résumé : OBJECTIVE: Several studies have explored the processing specificity of music and speech, but only a few have addressed the processing autonomy of their fundamental components: pitch and phonemes. Here, we examined the additivity of the mismatch negativity (MMN) indexing the early interactions between vowels and pitch when sung. METHODS: Event-related potentials (ERPs) were recorded while participants heard frequent sung vowels and rare stimuli deviating in pitch only, in vowel only, or in both pitch and vowel. The task was to watch a silent movie while ignoring the sounds. RESULTS: All three types of deviants elicited both an MMN and a P3a ERP component. The observed MMNs were of similar amplitude for the three types of deviants and the P3a was larger for double deviants. The MMNs to deviance in vowel and deviance in pitch were not additive. CONCLUSIONS: The underadditivity of the MMN responses suggests that vowel and pitch differences are processed by interacting neural networks. SIGNIFICANCE: The results indicate that vowel and pitch are processed as integrated units, even at a pre-attentive level. Music-processing specificity thus rests on more complex dimensions of music and speech.