Can we track the progression of Alzheimer's Disease via lexical-semantic variables in connected speech?

Abstract

Background

Alzheimer's disease (AD) is one of the most common neurodegenerative disorders worldwide and is characterized by problems with cognition and language, especially word-finding difficulties. The present study focuses on lexical-semantic features via five discourse variables reflecting word-finding difficulties, namely indefinite terms, lexical frequency, repetitions, semantic paraphasias, and use of pronouns. Our aim is twofold: testing whether these variables can discriminate healthy aging from AD, but also mild from moderate AD.

Method

105 participants were examined from the existing Pitt corpus (available on DementiaBank), which includes the Cookie Theft Picture Description task. 40 participants were healthy controls, 25 were mild AD participants, and 40 moderate AD participants.

Results

The moderate AD group differed significantly from healthy controls in terms of indefinite terms, repetitions, semantic paraphasias, and pronouns. For the latter variable, mild AD patients also differed significantly from healthy controls. However, none of the variables could differentiate mild from moderate AD.

Conclusion

Four out of five discourse variables could discriminate healthy aging from moderate AD, while only one could discriminate mild AD patients. This is therefore questioning current literature on connected-speech measures in AD and calling for further research on the variables that could better distinguish mild to moderate AD.