"More than a century after Wernicke and Broca established that speech perception and production rely on temporal and prefrontal cortices of the left brain hemisphere, the biological determinants for this organization are still unknown. While functional neuroanatomy has been described in great detail, the neuroscience of language still lacks a physiologically plausible model of the neuro-computational mechanisms for coding and decoding of speech acoustic signal. We propose to fill this gap by testing the biological validity and exploring the computational implications of one promising proposal, the Asymmetric Sampling in Time theory. AST assumes that speech signals are analysed in parallel at multiple timescales and that these timescales differ between left and right cerebral hemispheres. This theory is original and provocative as it implies that a single computational difference, distinct integration windows in right and left auditory cortices could be sufficient to explain why speech is preferentially processed by the left brain, and possible even why the human brain has evolved toward such an asymmetric functional organization. Our proposal has four goals: 1/ to validate, invalidate or amend AST on the basis of physiological experiments in healthy human subjects including functional magnetic resonance imaging (fMRI), combined electroencephalography (EEG) and fMRI, magnetoencephalography (MEG) and subdural electrocorticography (EcoG), 2/ to use computational modeling to probe those aspects of the theory that currently remain inaccessible to empirical testing (evaluation, assessment), 3/ to apply AST to binaural artificial hearing with cochlear implants, 4/ to test for disorders of auditory sampling in autism and dyslexia, two language neurodevelopmental pathologies in which a genetic basis implicates the physiological underpinnings of AST, and 5/ to assess potential generalisation of AST to linguistic action in the context of speech production."