Linguistic correlates of cognitive decline: collection and analysis of a spoken corpus of patients with dementia living in Basilicata

Abstract

The study aims to profile the linguistic-communicative disabilities related to dementia's onset in Italian. To this purpose, we recruited 40 subjects residing in Basilicata, an administrative region of Southern Italy not previously represented in this type of study. The enrolled cohort is balanced by sex and age: 20 healthy subjects and 20 subjects affected by different types of dementia (i.e., Alzheimer's disease, mixed dementia, frontotemporal dementia, and vascular dementia). After neuropsychological evaluation, we acquired approximately 9 hours of semi-spontaneous speech. Then, the corpus was transcribed and annotated to automatically extract a rich set of linguistic markers. Through statistical inference tests, we outline a degradation of the phonetic-acoustic, morphosyntactic/lexical, semantic, and syntax skills related to the disease. Finally, the qualitative analysis allows for better characterizing the linguistic profile of dementia, highlighting other traits such as local/global coherence and cohesion deficits, anomia's compensatory strategies, and repetitiveness/topic iteration.