Sponsor: Georgia Zellou, Ph.D. Linguistics Alzheimer's Disease (AD) affects roughly 7 million people aged 65 and older in the United States. Language can be an index of cognitive status; subtle shifts in word-finding, syntax, coherence, and speech timing often emerge early and track with disease progression. Yet, the majority of prior work has examined picture description tasks. This study examined autobiographical speech from older adults in the UC Davis Alzheimer's Disease Research Center corpus, including 221 cognitively normal, 99 with mild cognitive impairment, and 49 with dementia. Speakers responded to three questions about their life (their hometown, wedding, trip). Audio files were transcribed by Whisper. We manually corrected and annotated the transcripts, generating a 'ground truth' dataset. Annotations focused on where false starts and fillers were occurring and correcting ASR errors, such as due to background noise, speakers' accents/ dialects. We predict that speakers whose speech has disfluencies such as lower speech rate, longer pauses, or a higher proportion of non-substantive filler words will exhibit higher incidence of Alzheimer's or other cognitive decline. Overall, this research project uses an extensive database of manually corrected transcripts to look for linguistic biomarkers of Alzheimer's Disease to better understand the disease and aid in early detection. Negative Affective Reactivity Decreases with Age in Rhesus Monkeys

Abstract profile. Full document pending author claim.