Abstract
Background: Connected speech and language (CSL) analysis can differentiate typically aging individuals from those with dementia by identifying subtle changes in language-dependent measures, such as lexical diversity and fluency. Prior studies have linked features like slower speech rates and increased silent pauses with Alzheimer's disease (AD) neuropathology, but the pause patterns in preclinical AD and their syntactic locations remain unclear. Objective: To investigate silent pauses as a potential early detection tool for AD in cognitively unimpaired individuals and compare their sensitivity to traditional neuropsychological tests in distinguishing biomarker-positive and biomarker-negative groups. Methods: Cognitively unimpaired participants from the Wisconsin Registry for Alzheimer's Prevention with amyloid (A) and tau (T) PET imaging were included. Twenty-six participants with A+/T+ profiles were matched 1:1 with A-/T- individuals using propensity score matching for age, sex, literacy, and race (n = 52). Participants were majority female (75.6%) and non-Hispanic white (91%) with a mean (sd) age of 67.5 (6.2). CSL samples were derived from the Cookie Theft picture description task closest to PET scans. Pauses were annotated using ELAN at utterance-initial, clause-initial, and within-clause locations, with further categorization of within-clause pauses by part of speech (nouns, verbs, adjectives/adverbs). Pause rates and durations were analyzed using Mann-Whitney U tests, with effect sizes measured by Cliff's delta (δ). Results: A+/T+ individuals had longer adjective/adverb-initial pauses than A-/T- ( p = 0.016; δ = 0.39), outperforming the preclinical Alzheimer's cognitive composite ( p = 0.347, δ = 0.15). Conclusions: Silent pauses, especially adjective/adverb-initial, show promise as an early indicator of AD neuropathology. Replication in diverse cohorts with longitudinal analyses is warranted.
Funding Information
  • National Institute on Aging (R01 AG021155)
  • National Institute on Aging (R01 AG027161)
  • National Institute on Aging (R01 AG054059)
  • National Institute on Aging (R01 AG062285)
  • National Institute on Aging (R01 AG082052)

This publication has 76 references indexed in Scilit: