Formulaic Language in People with Probable Alzheimer's Disease: A Frequency-Based Approach.

Background: Language change can be a valuable biological marker of overall cognitive change in Alzheimer's disease (AD) and other forms of dementia. Previous reports have described increased use of language formulas in AD, i.e., combinations likely processed in a holistic manner. Words that commonly occur together are more likely to become a formula.

Objective: To determine if frequency of co-occurrence as one indicator for formulaic language can distinguish people with probable AD from controls and if variables are sensitive to time post-symptom onset.

Methods: We developed the Frequency in Language Analysis Tool (FLAT), which indicates degrees of formulaicity in an individual language sample. The FLAT accomplishes this by comparing individual language samples to co-occurrence data from the British National Corpus (BNC). Our analysis also contained more conventional language variables in order to assess novel contributions of the FLAT. We analyzed data from the Pitt Corpus, which is part of DementiaBank.

Results: Both conventional and co-occurrence variables were able to distinguish AD and control groups. According to co-occurrence data, people with probable AD produced more formulaic language than controls. Only co-occurrence variables correlated with disease progression.

Discussion: Frequency of word co-occurrences is one indicator for formulaicity and a valuable contribution to characterizing language change in AD.