Age Effects on Semantic Coherence: Latent Semantic Analysis Applied to Letter Fluency Data

Abstract

We investigated age-related changes in the semantic distance between successively generated words in two letter fluency tasks differing with respect to demands placed on executive control. The semantic distance was measured by Latent Semantic Analysis (LSA). The results show that older people have a larger semantic distance between successively generated items than young people, and that this effect is particularly pronounced in the more demanding fluency task. Taken together, our findings support the idea that elderly have a less distinct semantic network compared to young people while also demonstrating the feasibility of LSA as a powerful tool for delineating multifaceted aspects of semantic organization inherent in behavioural data from language production tasks.