A System for Computerized Analysis of Verbal Fluency Tests

I present VF-Clust, a system for computerized analysis of psychological tests of verbal fluency, which are used in clinical settings to detect and assess neurological pathologies such as Alzheimer’s disease. While the simple scoring of such tests may not be sensitive to underlying pathologies, more nuanced clustering analyses often are. Currently, clustering analyses on verbal fluency tests are conducted manually, by trained scorers, in a process that is labor-intensive and prone to human error and interrater variability. VF-Clust is a resource for automatically generating clustering analyses, on both semantic and phonemic verbal fluency test responses, by utilizing latent semantic analysis and computational methods for determining phonetic similarity. Results from a pilot study indicate that VF-Clust’s automatically determined clustering measures are as useful as their manually determined equivalents. Additionally, in this study repetitions were more numerous in verbal fluency test responses from patients with dementia from Alzheimer’s disease. In the case of phonemic verbal fluency, controls differed with patients with Alzheimer’s diagnoses on the number of words repeated, but not on the number of words generated, which is the standard measure by which the test is scored. While repetitions are typically ignored in even the most in-depth analyses of verbal fluency test responses, these results indicate that future analyses should consider them. VF-Clust is freely available upon request.