Lindsay, H., Linz, N., Troeger, J., & Alexandersson, J. (2019).
Published in: Proceedings of the 3rd International Conference on Natural Language and Speech Processing, 17–24.
Phonemic Verbal Fluency (PVF) is a cognitive assessment task where a patient is asked to produce words constrained to a given alphabetical letter for a specified time duration. Patient productions are later evaluated based on strategies to reveal crucial diagnostic information by manually scoring results according to predetermined clinical criteria. In this paper, we propose four alternative similarity metrics and evaluate them in a two-fold argument, using the clinical criteria as a baseline. First, we consider the capacity of each metric to model PVF production using a rank-based approach, and then consider the metrics ability to compute finer resolution clinical measures that are indicative of the underlying strategy. Automation of the clinical criteria and proposed metrics are evaluated on PVF performances for 16 letters from 32 healthy German students (n=512). Weighted phonemic edit distance performed best overall for modelling both production and strategy