Where do disfluencies come from? Towards a mechanistic account

<p>Naturalistic speech is strewn with disfluencies, such as repetitions (the the ball), filled pauses(the um ball), or self-corrections (the pall..ball). Understanding where these phenomena comefrom is important because disfluencies provide a window into the interaction between speechplanning, execution, and monitoring, while also influencing how interlocutors interpretspeakers and their messages. Focusing on neurotypical adult speech production, this reviewexamines accounts that attribute disfluencies to errors, planning delays, or communicativesignaling, together with evidence linking difficulties at different production stages to partlydistinct but overlapping disfluency patterns. We then propose MONET (MONitoring for Errorand Time), a mechanistic framework that explains how the interplay between speechplanning, execution, and monitoring may lead speakers to pause, repeat, interrupt, or repairtheir speech. We outline its empirical grounding, derive testable predictions, and identifydirections for computational implementation.</p>