The localization of cortical sites essential for language was assessed by stimulation mapping in the left, dominant hemispheres of 117 patients. Sites were related to language when stimulation at a current below the threshold for afterdischarge evoked repeated statistically significant errors in object naming. The language center was highly localized in many patients to form several mosaics of 1 to 2 sq cm, usually one in the frontal and one or more in the temporoparietal lobe. The area of individual mosaics, and the total area related to language was usually much smaller than the traditional Broca-Wernicke areas. There was substantial individual variability in the exact location of language function, some of which correlated with the patient's sex and verbal intelligence. These features were present for patients as young as 4 years and as old as 80 years, and for those with lesions acquired in early life or adulthood. These findings indicate a need for revision of the classical model of language localization. The combination of discrete localization in individual patients but substantial individual variability between patients also has major clinical implications for cortical resections of the dominant hemisphere, for it means that language cannot be reliably localized on anatomic criteria alone. A maximal resection with minimal risk of postoperative aphasia requires individual localization of language with a technique like stimulation mapping.

A model for the organization of language in the adult humans brain is derived from electrical stimulation mapping of several language-related functions: naming, reading, short-term verbal memory, mimicry of orofacial movements, and phoneme identification during neurosurgical operations under local anesthesia. A common peri-Sylvian cortex for motor and language functions is identified in the language dominant hemisphere, including sites common to sequencing of movements and identification of phonemes that may represent an anatomic substrate for the “motor theory of speech perception.” This is surrounded by sites related to short-term verbal memory, with sites specialized for such language functions as naming or syntax at the interface between these motor and memory areas. Language functions are discretely and differentially localized in association cortex, including some differential localization of the same function, naming, in multiple languages. There is substantial individual variability in the exact location of sites related to a particular function, a variability which can be partly related to the patient's sex and overall language ability and which may depend on prior brain injury and, perhaps subtly, on prior experience. A common “specific alerting response” mechanism for motor and language functions is identified in the lateral thalamus of the language–dominant hemisphere, a mechanism that may select the cortical areas appropriate for a particular language function.

Although non-invasive methods such as functional magnetic resonance imaging, electroencephalograms and magnetoencephalograms provide most of the current data about the human brain, their resolution is insufficient to show physiological processes at the cellular level. Clinical approaches sometimes allow invasive recordings to be taken from the human brain, mainly in patients with epilepsy or with movement disorders, and such recordings can sample neural activity at spatial scales ranging from single cells to distributed cell assemblies. In addition to their clinical relevance, these recordings can provide unique insights into brain functions such as movement control, perception, memory, language and even consciousness.