The neuronal infrastructure of speaking

Models of speaking distinguish producing meaning, words and syntax as three different linguistic components of speaking. Nevertheless, little is known about the brain’s integrated neuronal infrastructure for speech production. We investigated semantic, lexical and syntactic aspects of speaking using fMRI. In a picture description task, we manipulated repetition of sentence meaning, words, and syntax separately. By investigating brain areas showing response adaptation to repetition of each of these sentence properties, we disentangle the neuronal infrastructure for these processes. We demonstrate that semantic, lexical and syntactic processes are carried out in partly overlapping and partly distinct brain networks and show that the classic left-hemispheric dominance for language is present for syntax but not semantics.

► Meaning, words and syntax in the speaking brain are partly joint and partly separated.
► Left pMTG interfaces between these processes, and is involved in the lexicon.
► Syntax is left lateralized while meaning relies on bilateral contributions.
► It is feasible to investigate overt sentence-level speech in an fMRI environment.