Do Disfluencies Increase With Age? Evidence From a Sequential Corpus Study of Disfluencies

Speech disfluencies such as repeated words and pauses provide information about the cognitive systems underlying speech production. Understanding whether older age leads to changes in speech fluency can therefore help characterize the robustness of these systems over the life span. Older adults have been assumed to be more disfluent, but current evidence is minimal and contradictory. Particularly noteworthy is the lack of longitudinal data that would help establish whether a given individual's disfluency rates change over time. This study examines changes in disfluency rates through a sequential design with a longitudinal component, involving the analysis of 325 recorded interviews conducted with 91 individuals at several points in their lives, spanning the ages of 20-94 years. We analyzed the speech of these individuals to assess the extent to which they became more disfluent in later interviews. We found that, with older age, individuals spoke more slowly and repeated more words. However, older age was not associated with other types of disfluencies such as filled pauses (uh's and um's) and repairs. Overall, this study provides evidence that, although age itself is not a strong predictor of disfluencies, age leads to changes in other speech characteristics among some individuals (i.e., speech rate and indicators of lexical and syntactic complexity), and those changes in turn predict the production of disfluencies over the life span. These findings help resolve previous inconsistencies in this literature and set the stage for future experimental work on the cognitive mechanisms underlying changes in speech production in healthy aging. Public Significance Statement This study measures changes in speech fluency over the life span sequentially using a novel corpus of 91 individuals interviewed at multiple times in their lives. We found that, while older adults spoke more slowly and repeated more words, they were not more likely to produce filled pauses (uh's and um's) and repairs. Overall, we suggest that age is not as strong a predictor of disfluencies as are other individual differences in speech characteristics.