Beyond Accuracy: Cognitive Speed in Speech-in-Noise Processing and Its Implications for the Ease of Language Understanding Model.

Abstract

Purpose: The ease of language understanding (ELU) model predicts two processing streams-fast, implicit prediction and slower, explicit postdiction-challenging single-factor "general speed" accounts. The present study examined whether cognitive speed is unitary or fractionated, whether both speed and accuracy predict speech in noise (SPIN), and how SPIN relates to rapid automatized naming (RAN).

Method: Adults with normal hearing or hearing loss (n = 303) from the N200 study completed latency-based tasks indexing long-term memory access, executive control, Working Memory Tests, SPIN, and RAN. Exploratory and confirmatory factor analyses were used.

Results: Two speed factors emerged-long-term memory access speed and executive speed-contradicting a general speed model. Speed-SPIN associations were small and clearest in easier listening conditions. Executive speed predicted RAN. SPIN correlated with RAN, especially among hard of hearing participants.

Conclusions: Cognitive speed fractionates in line with the ELU. Working memory speed, unlike accuracy, does not drive SPIN. Executive speed selectively predicts RAN, and SPIN relates to RAN, supporting SPIN's potential as an early proxy for cognitive decline.