Comparison of category and letter fluency tasks through automated analysis

Area of Science:

• Neuroscience
• Linguistics
• Computational Linguistics

Background:

• Category and letter fluency tasks are standard neuropsychological tools for assessing lexical retrieval.
• Previous research has primarily relied on manual analysis, limiting the scope of investigation.

Purpose of the Study:

• To analyze speech production in category and letter fluency tasks using validated automated methods.
• To compare lexical and acoustic measures between category and letter fluency tasks in healthy adults.

Main Methods:

• Healthy participants (n=36) completed online category ("Animal") and letter ("F") fluency tasks.
• Automated pipelines utilizing natural language processing and acoustic analysis processed recordings.
• Lexical variables (e.g., age of acquisition, ambiguity, concreteness, frequency, familiarity) and acoustic measures (e.g., inter-word response time, word duration) were analyzed.

Main Results:

• Category fluency yielded significantly more words than letter fluency.
• Category fluency was associated with lower age of acquisition, frequency, semantic ambiguity, semantic distance, and inter-word response time, but higher concreteness and familiarity compared to letter fluency.
• Word characteristics like familiarity, semantic ambiguity, semantic distance, and duration changed over time and interacted significantly between tasks.

Conclusions:

• Lexical measures (age of acquisition, familiarity, semantic ambiguity) and acoustic measures (inter-word response time, word duration) are crucial for differentiating category and letter fluency tasks.
• Reproducible and scalable automated techniques can efficiently detect differences in fluency tasks.
• These automated methods hold promise for expanding understanding of speech feature differences in clinical populations.