Typical aging is characterized by an increase of tip of the tongue experiences. These word finding difficulties reflect an inability to retrieve a word despite a strong feeling of knowing it. Unlike typical aging, word finding difficulties are due to a lexical-semantic impairment in Alzheimer’s disease. Even though these word retrieval failures are not similar, many authors analyse disfluencies during discourse production (e.g., pauses, repetitions, self-corrections) as a proof of word finding difficulties, in both populations. That is why this project first aims at characterizing how to tackle word finding difficulties due to typical aging, using three types of discourse tasks. I will then apply these findings to distinguish typical aging from Alzheimer’s disease, using the TalkBank database. Moreover, because the verbal monitoring system could help resolve a lexical retrieval failure, the current project also aims at better characterizing this system. It will unravel how verbal monitoring is represented in the brain, by re-analysing data from an fMRI tongue twister task. Based on these findings, I will test how it affects typical aging, Alzheimer’s disease, and word finding difficulties. This project has several theoretical implications for research on aging, monitoring, and disfluencies. Because discourse tasks are becoming more and more used in clinical settings, it will also open new directions for the assessment of patients’ communication difficulties in daily life.