Essentials: The Science of Learning & Speaking Languages | Dr. Eddie Chang

Dr. Eddie Chang, a neurosurgeon at UCSF, joins Andrew Huberman to explore the neuroscience of speech and language. He begins by distinguishing the two: speech refers to the physical motor act of producing sound through the vocal tract, while language encompasses broader cognitive functions including semantics (meaning), syntax (grammar), and pragmatics (contextual understanding). Chang explains the mechanics of vocalization — how the larynx brings vocal folds together during exhalation to produce voicing around 100 Hz in men and 200 Hz in women, and how the pharynx, oral cavity, tongue, and lips then shape that raw sound into intelligible consonants and vowels. He notes that primitive vocalizations like crying and moaning are governed by evolutionarily older brain areas distinct from speech and language regions, which is why brainstem stroke patients can still moan but not speak.

The conversation then moves to Chang's clinical research on brain-machine interfaces (BMI) for paralyzed patients. He describes conditions like brainstem stroke and ALS that result in 'locked-in syndrome,' where cognitive function is fully intact but voluntary movement and speech are impossible. Chang describes the BRAVO clinical trial, in which electrodes were implanted over speech motor cortex areas in a patient named Pancho — a man paralyzed for 15 years following a brainstem stroke after a car accident. The system intercepts cortical signals as the patient attempts to speak, converts them via machine learning algorithms into decoded words, and uses autocorrect-style language models (trained on 50-word vocabularies) to improve accuracy. This represented the first time a paralyzed individual's intended speech was decoded directly from brain activity into words and sentences.

Chang also discusses the future of this technology, including the development of computer-animated avatars that can reproduce not only decoded speech but also facial expressions and mouth movements, making communication more natural and holistic. He envisions a near-term future where social interactions increasingly occur in digital/virtual spaces, making such avatar-based communication tools essential for disabled individuals. He connects this to broader consumer trends, suggesting that avatar-speech will become a mainstream communication format.

On the topic of cognitive augmentation via neurotechnology, Chang acknowledges that the field is progressing but argues that current commercial technologies are far from matching the bandwidth and sophistication of naturally evolved neural speech systems. He raises important ethical concerns about who gets access to augmentation technologies, whether society has thought through the implications, and notes that augmentation itself is not a new human behavior — caffeine, medications, and smartphones already augment cognition.

Finally, Chang discusses stuttering as a speech (not language) condition involving a breakdown in the precise motor coordination required for fluent speech. He explains that anxiety can exacerbate stuttering but does not cause it, and that auditory feedback — hearing oneself speak — plays a critical role, offering clues about the neural mechanisms involved. Treatment typically involves speech therapy focused on creating conditions for fluent initiation, and early intervention during periods of high neuroplasticity is generally beneficial.