Selection of the Most Suitable Linguistic Features for Diagnosis of Alzheimer’s Disease from Changes in Spoken Language Production

Abstract: Alzheimer’s Disease (AD) is one of the most widespread, neurodegenerative diseases in the world. It affects millions of people; and although it is widely researched, an efficient cure has not been discovered yet. Identification of the disease is often costly and time consuming. Additionally, the used tests only allow us to detect AD once it is in progress. Some longitudinal studies have suggested that there is a possibility that changes in certain linguistic features could help us identify future AD patients from their written language production before other changes can be detected. Other studies have identified linguistic differences between people diagnosed with AD and AD-free controls. This paper aimed to find out whether a longitudinal study on oral language production will follow the same patterns as suggested by previous studies that longitudinally examined written production; and studies that cross-sectionally examined oral production. Changes in several linguistic features – number of unique types, frequency of verbs, frequency of adverbs, frequency of coordinate conjunctions, pronoun/noun ratio, number of filled pauses, and number of cohesivity interruptions – were observed in data collected from interviews of several famous people that were later in life diagnosed with AD. The collected data was analysed with simple linear regression. Number of unique types and number of cohesivity interruptions were recognised as the features with the best potential for future studies and their possible use was described.