Revisiting human language and speech production network: A meta-analytic connectivity modeling study

Abstract

In recent decades, converging evidence has reached a consensus that human speech production is carried out by large-scale hierarchical network comprising both language-selective and domain-general systems. However, it remains unclear how these systems interact during speech production and the specific contributions of their component regions. By utilizing a series of meta-analytic approaches based on various language tasks, we dissociated four major systems in this study: domain-general, high-level language, motor-perception, and speech-control systems. Using meta-analytic connectivity modeling, we found that while the domain-general system is coactivated with high-level language regions and speech-control networks, only the speech-control network at the ventral precentral gyrus is coactivated with other systems during different speech-related tasks, including motor perception. In summary, this study revisits the previously proposed language models using meta-analytic approaches and highlights the contribution of the speech-control network to the process of speech production independent of articulatory motor.