A study published in Nature (DOI: 10.1038/s41586-023-06982-w) reveals that neurons in the language-dominant prefrontal cortex of humans encode detailed information about the phonetic structure of planned words during speech production. Using acute ultrahigh-density Neuropixels recordings, the researchers identified neurons that represent the specific order and structure of articulatory events before speech, as well as the segmentation of phonetic sequences into syllables. These neurons accurately predict the phonetic, syllabic, and morphological components of upcoming words and show a temporally ordered dynamic. The findings suggest a structured organization of phonetic representations by prefrontal neurons, demonstrating a cellular process that supports speech production. The study shows that these neurons are broadly organized along the cortical column and their activity patterns transition from articulation planning to production. They also reliably track the detailed composition of consonant and vowel sounds during perception and distinguish processes related to speaking from those related to listening. The neurons encode information about the phonetic composition of upcoming words, including specific syllables and morphemes, and their activity patterns are significantly informative of multiple features. The study also reveals that the activities of these neurons are temporally structured, with decoding performance peaking for morphemes first, followed by phonemes and syllables. The neurons show a transition from articulation planning to production, with their activities reflecting the planned phonemes and syllables. The findings suggest that these neurons occupy distinct functional subspaces during planning and production, allowing for the stable representation of phonetic, syllabic, and morphological contents of words. The study highlights the importance of the prefrontal cortex in speech production and provides insights into the cellular mechanisms underlying speech. It also supports the existence of context-general representations of speech sounds and suggests that the prefrontal cortex may be involved in a 'mirror' system that allows for the shared representation of phonetic features during speaking and listening. The findings have implications for understanding speech production and could inform the development of synthetic speech prostheses and other brain-machine interface technologies.A study published in Nature (DOI: 10.1038/s41586-023-06982-w) reveals that neurons in the language-dominant prefrontal cortex of humans encode detailed information about the phonetic structure of planned words during speech production. Using acute ultrahigh-density Neuropixels recordings, the researchers identified neurons that represent the specific order and structure of articulatory events before speech, as well as the segmentation of phonetic sequences into syllables. These neurons accurately predict the phonetic, syllabic, and morphological components of upcoming words and show a temporally ordered dynamic. The findings suggest a structured organization of phonetic representations by prefrontal neurons, demonstrating a cellular process that supports speech production. The study shows that these neurons are broadly organized along the cortical column and their activity patterns transition from articulation planning to production. They also reliably track the detailed composition of consonant and vowel sounds during perception and distinguish processes related to speaking from those related to listening. The neurons encode information about the phonetic composition of upcoming words, including specific syllables and morphemes, and their activity patterns are significantly informative of multiple features. The study also reveals that the activities of these neurons are temporally structured, with decoding performance peaking for morphemes first, followed by phonemes and syllables. The neurons show a transition from articulation planning to production, with their activities reflecting the planned phonemes and syllables. The findings suggest that these neurons occupy distinct functional subspaces during planning and production, allowing for the stable representation of phonetic, syllabic, and morphological contents of words. The study highlights the importance of the prefrontal cortex in speech production and provides insights into the cellular mechanisms underlying speech. It also supports the existence of context-general representations of speech sounds and suggests that the prefrontal cortex may be involved in a 'mirror' system that allows for the shared representation of phonetic features during speaking and listening. The findings have implications for understanding speech production and could inform the development of synthetic speech prostheses and other brain-machine interface technologies.