Short speech tasks can reveal cognitive changes without requiring a long clinical interview. This paper examines whether automatic speech recognition can evaluate immediate repetition of nonsense words and distinguish healthy participants from people with cognitive impairment. The experiment shows why recognition errors, phonological similarity and the choice of language model all matter when speech technology becomes part of a screening method.
Repeating nonsense words is a compact task with a surprisingly rich signal. It exercises speech perception, short-term memory and speech production while avoiding the advantage of simply knowing a familiar word.
The paper studies how automatic speech recognition can support analysis of this task in cognitive-disorder research. It is part of the DigiDiaDem line of work, where speech and language technology is used to build measurable, repeatable components for remote speech-and-memory assessment.
The task complements the broader DigiDiaDem Speech-Cognitive Dataset study and the related work on semantic analysis of spoken image descriptions.
Authors
Jan Tupý, Jan Švec, Luboš Šmídl
Abstract
We investigate whether immediate repetition of nonsense words can distinguish cognitively healthy adults from those with mild cognitive impairment (MCI) or dementia. In a computer-based study, 129 Czech speakers (45–84 y) repeated six pseudowords; each session was recorded and transcribed by four state-of-the-art ASR models. Grapheme-level similarity between the transcript and the target word served as a phonological accuracy score. Using logistic regression, the best ASR variant (wav2vec-nolm, no language model) separated patients from controls with 77 % accuracy, whereas language-model–’corrected’ hypotheses performed markedly worse. These findings show that a very short, vocabulary-free task can reveal early linguistic decline while requiring only consumer hardware and minimal instruction. Incorporating such a nonsense word repetition subtask into digital neuropsychological batteries could therefore sharpen large-scale, remote screening and help clinicians focus full assessments on individuals at greatest risk.